UK Vintage Radio Repair and Restoration Discussion Forum

UK Vintage Radio Repair and Restoration Discussion Forum (https://www.vintage-radio.net/forum/index.php)
-   Homebrew Equipment (https://www.vintage-radio.net/forum/forumdisplay.php?f=99)
-   -   Advice on this linestage build (https://www.vintage-radio.net/forum/showthread.php?t=180502)

nzoomed 24th May 2021 10:48 am

Advice on this linestage build
 
1 Attachment(s)
I've put this together from different sources.

I basically want a buffered line stage with minimal gain since most digital sources today are able to drive a power amp fine, but may still require a little gain for the phono, I will have an RIAA stage fed into this for the phono input, but may need some gain depending on how much gain I get from the 6922/E88CC that it uses.

I have built on the Baxandall design found here:
https://www.angelfire.com/electronic...mp-Tone-A.html
https://www.angelfire.com/electronic...e-fig-06-1.gif

It uses a 12AU7 as a cathode follower for the input, I've modified it by adding an EF86 in triode mode at the end, which should give a modest gain, but make up for the losses in the tone stack.
The EF86 configuration is essentially based on the Svetlana preamp schematic.

I have also added balance controls and a tone bypass. I'm not 100% sure how this will behave.
Im also interested if there is a way to add NFB to the EF86 to reduce gain and improve linearity like ive read a bit about, I might be able to incorporate a switch into this, or utilize the tone bypass switch to do this, as its possible the volume/gain may increase when bypassed?

Robert Gribnau 24th May 2021 2:33 pm

Re: Advice on this linestage build
 
The second valve misses a grid resistor.

I am afraid that your tone control is not going to work very effectively because the gain of the second valve is only about 20. That gain is used (by feedback) for the tone control. A gain of only 20 is too low to cover the range of the tone controls from the lowest to the highest settings (about - 15 dB to + 15 dB). In the schematic you linked to, the gain of the second valve is about 70.

nzoomed 25th May 2021 9:11 am

Re: Advice on this linestage build
 
Quote:

Originally Posted by Robert Gribnau (Post 1377381)
The second valve misses a grid resistor.

I am afraid that your tone control is not going to work very effectively because the gain of the second valve is only about 20. That gain is used (by feedback) for the tone control. A gain of only 20 is too low to cover the range of the tone controls from the lowest to the highest settings (about - 15 dB to + 15 dB). In the schematic you linked to, the gain of the second valve is about 70.

I wasnt sure if a grid resistor was needed or not on the second tube, the first one didnt have one.

Anyway, as far as gain goes, I thought 25 was going to be too much, If i configure this as a pentode will I be any better off?

Robert Gribnau 25th May 2021 9:39 am

Re: Advice on this linestage build
 
The first valve does have a grid resistor (470K).

With the EF86 in pentode mode the gain will be about 110.

To be honest: I am not sure about the effect of the high output impedance of the second valve in your circuit (which will be even higher with the EF86 in pentode mode) on the way the controls will work (will the frequency curves be the same?). In the schematic you linked to, the second valve is followed by a cathode-follower, which has low output impedance (and a 'gain' of a little under 1).

nzoomed 25th May 2021 11:33 am

Re: Advice on this linestage build
 
Quote:

Originally Posted by Robert Gribnau (Post 1377596)
The first valve does have a grid resistor (470K).

With the EF86 in pentode mode the gain will be about 110.

To be honest: I am not sure about the effect of the high output impedance of the second valve in your circuit (which will be even higher with the EF86 in pentode mode) on the way the controls will work (will the frequency curves be the same?). In the schematic you linked to, the second valve is followed by a cathode-follower, which has low output impedance (and a 'gain' of a little under 1).

Sorry, I was referring to the first schematic, neither of the last two stages have any grid resistors, on either the 12au7 or 12ax7.

I was hoping I could have reduced the tube count by using an EF86 in this way, plus i have a ton of the things so thought it would be good to make use of them.

I have been reading this here and it shows it in a pentode configuration.
It says the EF86 has alot of gain so has used an EF184. I have a few of these too.
https://www.sciencedirect.com/topics...ring/baxandall

Would I be better off keeping the EF86 in triode mode and adding a 12AU7 cathode follower at the end?
From what i understand, a 12AU7 can take a high impedance load and give a low impedance output which is useful in a line stage.

Robert Gribnau 25th May 2021 12:02 pm

Re: Advice on this linestage build
 
I am not sure what you mean by the first schematic. If it is the schematic of the tweaked Brimar tone control, than the 12AX7 section has its grid returned to ground through the 470K and 2.2 M resistors. The grid of the 12AU7 section following the 12AX7 is directly coupled to the anode of the 12AX7 and therefore does not need a grid resistor.

Like I wrote, I think that the gain of an EF86 in triode mode is not high enough to cover the range from - 15 dB to + 15 dB. An EF184 in triode mode (mu is about 60 so the gain can be about 40) would be somewhat better. I have never used the EF184 before but I have read that the EF184 can be microphonic.

I hope some other forum members join in because the precise workings of tone controls is a little over my head (except in guitar amplifiers, I never build tone controls in my projects).

nzoomed 25th May 2021 12:17 pm

Re: Advice on this linestage build
 
Quote:

Originally Posted by Robert Gribnau (Post 1377639)
I am not sure what you mean by the first schematic. If it is the schematic of the tweaked Brimar tone control, than the 12AX7 section has its grid returned to ground through the 470K and 2.2 M resistors. The grid of the 12AU7 section following the 12AX7 is directly coupled to the anode of the 12AX7 and therefore does not need a grid resistor.

Like I wrote, I think that the gain of an EF86 in triode mode is not high enough to cover the range from - 15 dB to + 15 dB. An EF184 in triode mode (mu is about 60 so the gain can be about 40) would be somewhat better. I have never used the EF184 before but I have read that the EF184 can be microphonic.

I hope some other forum members join in because the precise workings of tone controls is a little over my head (except in guitar amplifiers, I never build tone controls in my projects).


It was this one here.
https://www.angelfire.com/electronic...e-fig-06-1.gif
I think you are looking at a different schematic.
There is a 1M from the grid to ground and perhaps thats what im missing?

I hear alot say the EF86 is microphonic too, im not sure if there is much difference between the two in regards to microphony.

Robert Gribnau 25th May 2021 1:33 pm

Re: Advice on this linestage build
 
The EF86 was designed for low level audio applications. It has a bifilar filament (like the EF40 and the CF50 before it) to minimize heater hum. With respect to noise and microphonics I think they are rather good. The EF86's I use in the first stage of two of my guitar amplifiers are not prone to microphonics (but they are stand-alone amplifiers and the filaments are being fed dc voltage).

The EF184 was designed for the IF stages of televisions, so also mostly for low level signals. But they were not designed with audio in mind (as far as I know, they do not have bifilar filaments). I know that Radford used them in their STA-100 amplifier as phase splitter but there they operate with high input signals.

The forerunner of the EF184 was the EF80. Some two years ago there was a discussion on this forum about its usability for audio: https://www.vintage-radio.net/forum/...d.php?t=157707

Robert Gribnau 25th May 2021 4:14 pm

Re: Advice on this linestage build
 
What you perhaps could do is use the schematic of the passive tone control on the site you linked to, preceded by an EF86 in triode mode + a 12AU7 cathode-follower (dc-coupled). The EF86 gives you a gain of about 20, so of about 26 dB, while the tone control takes 20 dB off of that. You than have about 6 dB headroom.

nzoomed 26th May 2021 1:05 am

Re: Advice on this linestage build
 
Quote:

Originally Posted by Robert Gribnau (Post 1377668)
The EF86 was designed for low level audio applications. It has a bifilar filament (like the EF40 and the CF50 before it) to minimize heater hum. With respect to noise and microphonics I think they are rather good. The EF86's I use in the first stage of two of my guitar amplifiers are not prone to microphonics (but they are stand-alone amplifiers and the filaments are being fed dc voltage).

The EF184 was designed for the IF stages of televisions, so also mostly for low level signals. But they were not designed with audio in mind (as far as I know, they do not have bifilar filaments). I know that Radford used them in their STA-100 amplifier as phase splitter but there they operate with high input signals.

The forerunner of the EF184 was the EF80. Some two years ago there was a discussion on this forum about its usability for audio: https://www.vintage-radio.net/forum/...d.php?t=157707

Interesting what you say about the EF184, I have no idea if it has a spiral heater or not.
I guess going to DC heaters could solve this?
I know both the EF80 and EF184 are popular tubes with builders.

Quote:

Originally Posted by Robert Gribnau (Post 1377728)
What you perhaps could do is use the schematic of the passive tone control on the site you linked to, preceded by an EF86 in triode mode + a 12AU7 cathode-follower (dc-coupled). The EF86 gives you a gain of about 20, so of about 26 dB, while the tone control takes 20 dB off of that. You than have about 6 dB headroom.

I was actually wondering about that myself.
I had also considered using it as the first gain stage, but I think impedance matching would be an issue?

From what I understand, with a line stage, particularly if you are using it as a buffer with little gain, is you want to keep output impedance as low as possible, but able to match various loads on the input?

Diabolical Artificer 26th May 2021 5:16 am

Re: Advice on this linestage build
 
Quote:

I know both the EF80 and EF184 are popular tubes with builders.
Mainly cos the EF80 is cheap and the EF184 cos it's mentioned quite often in thread's. The 184 comes up a lot I think cos it was used in a lot of scope front ends, it's also developed a self perpetuating self affirmation cult due to all the threads on it, not to say it's a "bad" valve.

That tone control circuit you attached is touchy, as regards gain as Robert says, I tried modding it using an ECF82 after the tone control to get extra gain in one valve base, no go. As per the author's schematic it works, mess with the gain and it's essentially an attenuator. In my case I added an ECC82 gain stage, 47k Ra, 1m grid leak. 1k ish Rk after the published circuit. You could use a CCDA line stage though.

Andy.

nzoomed 26th May 2021 9:09 am

Re: Advice on this linestage build
 
Quote:

Originally Posted by Diabolical Artificer (Post 1377869)
Quote:

I know both the EF80 and EF184 are popular tubes with builders.
Mainly cos the EF80 is cheap and the EF184 cos it's mentioned quite often in thread's. The 184 comes up a lot I think cos it was used in a lot of scope front ends, it's also developed a self perpetuating self affirmation cult due to all the threads on it, not to say it's a "bad" valve.

That tone control circuit you attached is touchy, as regards gain as Robert says, I tried modding it using an ECF82 after the tone control to get extra gain in one valve base, no go. As per the author's schematic it works, mess with the gain and it's essentially an attenuator. In my case I added an ECC82 gain stage, 47k Ra, 1m grid leak. 1k ish Rk after the published circuit. You could use a CCDA line stage though.

Andy.

OK, well I may just end up going with the schematic as it is.
Will adding balance controls or a tone bypass affect anything?

I was hoping to make use of the EF86 in my build just for the sake I have so many of the things.

If anyone has any similar schematics to share, that would be great.

Jez1234 26th May 2021 1:43 pm

Re: Advice on this linestage build
 
EF86 or better still EF184 in pentode form giving around x 400 gain followed by a cathode follower then wrap the feedback around the Baxandall network.

There are many ways of making a pre amp and hundreds of schematics on the web.

There will be plenty of designs from back in the day which you could just remove the phono section from. Beware of some old designs being intended to drive >100K load though! Keep the cathode follower.

If you didn't want tone controls I'd have linked to a design of my own intended for easy DIY build.

Robert Gribnau 26th May 2021 2:54 pm

Re: Advice on this linestage build
 
1 Attachment(s)
I put this together (my suggestion in post #9 but with a cathode-follower after the tone controls).

Jez1234 26th May 2021 6:36 pm

Re: Advice on this linestage build
 
FWIW after seeing reports of super linearity from EF86 when triode strapped I tried it....

It's not true! I tried several samples, tried different methods of triode strapping and experimented with load line etc and distortion was disappointing.

nzoomed 27th May 2021 1:33 am

Re: Advice on this linestage build
 
Quote:

Originally Posted by Jez1234 (Post 1377984)
EF86 or better still EF184 in pentode form giving around x 400 gain followed by a cathode follower then wrap the feedback around the Baxandall network.

There are many ways of making a pre amp and hundreds of schematics on the web.

There will be plenty of designs from back in the day which you could just remove the phono section from. Beware of some old designs being intended to drive >100K load though! Keep the cathode follower.

If you didn't want tone controls I'd have linked to a design of my own intended for easy DIY build.

That sort of sounds similar to what Robert has suggested in post #14 except his is not a baxandall feedback loop, I may be tempted to go with something like that and source some EF184 tubes perhaps.
How easy is it to implement a bypass switch on a baxandall design? Ignoring the rest of my schematic on the first post, would that way of inserting a switch work? Will the volume level change significantly when switchong on or off the tone stack?

Quote:

Originally Posted by Robert Gribnau (Post 1377997)
I put this together (my suggestion in post #9 but with a cathode-follower after the tone controls).

Thanks for that, this looks similar to what I was looking at to begin with, but then I got onto the baxandall unity gain design and thought that may have been a better option, but if i can make this work with a baxandall layout, ill probably go with this.

Quote:

Originally Posted by Jez1234 (Post 1378049)
FWIW after seeing reports of super linearity from EF86 when triode strapped I tried it....

It's not true! I tried several samples, tried different methods of triode strapping and experimented with load line etc and distortion was disappointing.


Thats interesting to hear, I have read from others experiences that if you feed negative feedback into a triode strapped EF86, that it improves linearity immensely.
Does that mean that an ef86 as a pentode has poor linearity?

Diabolical Artificer 27th May 2021 5:08 am

Re: Advice on this linestage build
 
Feeding an EF86with an active load improves linearity, just a MJE350 CCS. That said I don't mind a bit of distortion in my amps, else why use valves?

Andy.

Robert Gribnau 27th May 2021 5:08 am

Re: Advice on this linestage build
 
Hello Jez,

Quote:

Originally Posted by Jez1234 (Post 1377984)
EF86 or better still EF184 in pentode form giving around x 400 gain followed by a cathode follower then wrap the feedback around the Baxandall network.

Could you give values for Vb, Ra, Rg2 and Rk to achieve a gain of 400 with an EF86 and an EF184?

Quote:

Originally Posted by Jez1234 (Post 1378049)
FWIW after seeing reports of super linearity from EF86 when triode strapped I tried it....

It's not true! I tried several samples, tried different methods of triode strapping and experimented with load line etc and distortion was disappointing.

Can you share some of your test results with us? Did your results differ from the distortion figures in this datasheet (see page 3), and if so, in what way? https://frank.pocnet.net/sheets/030/e/EF86.pdf

Radio Wrangler 27th May 2021 5:28 am

Re: Advice on this linestage build
 
There seems to be some sort of cultural imperative driving people towards using triodes everywhere, but as Robert has pointed out, triodes don't have the necessary gain to make the active Baxendall tone control circuit work. Towards max boost it will run out of loop gain.

It's a feedback circuit and all the usual equations apply.

You could try a 2-stage amplifier, but then that doesn't invert so the feedback is the wrong sense. Alternatively a cascode would boost the gain.

Even with a pentode amplifying its little heart out, the gain is somewhat limited by the impedance of the feedback network loading the anode resistor. It's still not exactly wonderful, but nevertheless, a lot better than a triode alone.

What the circuit really needs is an NE5534. Don't worry, it won't pollute your sound. It's likely that some of what you listen to has already been through a load of 'em in the studio.

David

Jez1234 27th May 2021 12:22 pm

Re: Advice on this linestage build
 
Quote:

Originally Posted by Robert Gribnau (Post 1378150)
Hello Jez,

Quote:

Originally Posted by Jez1234 (Post 1377984)
EF86 or better still EF184 in pentode form giving around x 400 gain followed by a cathode follower then wrap the feedback around the Baxandall network.

Could you give values for Vb, Ra, Rg2 and Rk to achieve a gain of 400 with an EF86 and an EF184?

Quote:

Originally Posted by Jez1234 (Post 1378049)
FWIW after seeing reports of super linearity from EF86 when triode strapped I tried it....

It's not true! I tried several samples, tried different methods of triode strapping and experimented with load line etc and distortion was disappointing.


Can you share some of your test results with us? Did your results differ from the distortion figures in this datasheet (see page 3), and if so, in what way? https://frank.pocnet.net/sheets/030/e/EF86.pdf

I'm afraid I have better things to do than act as a free R & D service! There will be many designs already around. With active load you can get gains of thousands!

The distortion figures in that datasheet look about right yes and similar to what I obtained. There is a Svetlana datasheet suggesting 0.1% for something like 10V output!! No way. Suggestions of triode strapped EF86 being super linear abound on the internet also....

Jez1234 27th May 2021 12:48 pm

Re: Advice on this linestage build
 
Quote:

Originally Posted by nzoomed (Post 1378145)
Quote:

Originally Posted by Jez1234 (Post 1377984)
EF86 or better still EF184 in pentode form giving around x 400 gain followed by a cathode follower then wrap the feedback around the Baxandall network.

There are many ways of making a pre amp and hundreds of schematics on the web.

There will be plenty of designs from back in the day which you could just remove the phono section from. Beware of some old designs being intended to drive >100K load though! Keep the cathode follower.

If you didn't want tone controls I'd have linked to a design of my own intended for easy DIY build.

That sort of sounds similar to what Robert has suggested in post #14 except his is not a baxandall feedback loop, I may be tempted to go with something like that and source some EF184 tubes perhaps.
How easy is it to implement a bypass switch on a baxandall design? Ignoring the rest of my schematic on the first post, would that way of inserting a switch work? Will the volume level change significantly when switchong on or off the tone stack?

Quote:

Originally Posted by Robert Gribnau (Post 1377997)
I put this together (my suggestion in post #9 but with a cathode-follower after the tone controls).

Thanks for that, this looks similar to what I was looking at to begin with, but then I got onto the baxandall unity gain design and thought that may have been a better option, but if i can make this work with a baxandall layout, ill probably go with this.

Quote:

Originally Posted by Jez1234 (Post 1378049)
FWIW after seeing reports of super linearity from EF86 when triode strapped I tried it....

It's not true! I tried several samples, tried different methods of triode strapping and experimented with load line etc and distortion was disappointing.


Thats interesting to hear, I have read from others experiences that if you feed negative feedback into a triode strapped EF86, that it improves linearity immensely.
Does that mean that an ef86 as a pentode has poor linearity?

I had a look at your first link and right down the page is Fig 11.1 which is what I was suggesting. Check out how tone defeat is done in usual practice and it should work fine here also.
6AU6 is used as the pentode in that schematic but many types will do the job, ideally with changes to resistor values for optimum operation.

Negative feedback (NFB) improves linearity of any device at the expense of gain.
Pentodes are intrinsically less linear than triodes as triodes have "built in" NFB which results in their much lower gain and lower anode resistance.
Alternatively the high gain of pentodes can be used to give high loop gain (high NFB) for excellent overall linearity.

Robert Gribnau 27th May 2021 12:56 pm

Re: Advice on this linestage build
 
Hello Jez,

I did not ask you for free R & D service. I asked you because I could not believe you can achieve a gain of 400 with an EF86 or an EF184 at line level.

From your reaction I now understand that additional circuitry is needed. But even than I still find it hard to believe. But hey, I am just an amateur.

About the distortion of an EF86 strapped as a triode: I find it hard to understand that an EE in audio and a Chief engineer at a (now defunct) hi fi company, like you wrote you are, needs to prove to himself that those claims were false. Furthermore I do not know of any Svetlana datasheet that claimes 0,1 % at 10 V output for an EF86 in triode mode but if you say so, it must be out there somewhere.

Jez1234 27th May 2021 1:14 pm

Re: Advice on this linestage build
 
Quote:

Originally Posted by Diabolical Artificer (Post 1378149)
Feeding an EF86with an active load improves linearity, just a MJE350 CCS. That said I don't mind a bit of distortion in my amps, else why use valves?

Andy.

I like to try and rival SS THD figures with valves:)

Jez1234 27th May 2021 2:38 pm

Re: Advice on this linestage build
 
Quote:

Originally Posted by Robert Gribnau (Post 1378281)
Hello Jez,

I did not ask you for free R & D service. I asked you because I could not believe you can achieve a gain of 400 with an EF86 or an EF184 at line level.

From your reaction I now understand that additional circuitry is needed. But even than I still find it hard to believe. But hey, I am just an amateur.

About the distortion of an EF86 strapped as a triode: I find it hard to understand that an EE in audio and a Chief engineer at a (now defunct) hi fi company, like you wrote you are, needs to prove to himself that those claims were false. Furthermore I do not know of any Svetlana datasheet that claimes 0,1 % at 10 V output for an EF86 in triode mode but if you say so, it must be out there somewhere.

Well lookie here... seems I have my first troll on vintage radio.net:-)

It is easy to achieve gain of 400 from many pentodes. No "additional circuitry" is needed. Just a high value anode resistor and no appreciable load on it.

The Svetlana web site seems to be defunct. They are now made by a different supplier and just branded Svetlana and it seems they have gone to quite some lengths to remove all links to original datasheets etc and the current datasheet I can find is different from the original one... however here is a link to the relevant information, originally on page two of the datasheet http://users.telenet.be/tony.de.lobelle/toon/

<0.05% @ 10V RMS actually claimed.

Robert Gribnau 29th May 2021 7:45 am

Re: Advice on this linestage build
 
I stand corrected concerning the Svetlana claim.

My question about the gain of 400 from an EF86 and an EF184 at line level remains unanswered. I did not ask you about "many pentodes".

Calling forum members a troll is something you seem to do more often (on other forums and now also on this forum). My posting (# 22) was probably out of line. But it was caused by you asking for money to prove your statement (while we are in the "homebrew equipment" section).

Radio Wrangler 29th May 2021 12:08 pm

Re: Advice on this linestage build
 
A quick look at the anode characteristic of the EF86 in pentode mode and low current shows an anode curve slope of about 2.4 Megohms can be achieved. With a Gm of 1.8mA/v this would imply a gain of 4320 would be available with a perfect current source biasing the valve, and an infinite impedance load.

This is clearly silly.

At the low current for the flat anode characteristic, Gm will be appreciably reduced to maybe 0.5mA/V so that gain is down to 1200.

But this is still assuming there is no load on the thing and so it is doing no useful work. The Zin of the following stage will bring the gain crashing down 220k will give us about 100 gain.

If the impedance is left high enough to get more gain, the effect of stray capacitance will bring the -3dB frequency point crashing down. Heaven help us if it's driving a bit of screened wire to apower amp, or a Baxendall tone control feedback network.

So, yes, very high gains are possible from an EF86, but under such impractical conditions that you can't really use them.

There is also the issue of the control of the anode voltage with constant-current source anode bias. There will be contention with the large anode impedance of the valve alone, and very poor control of the quiescent anode voltage.

David

merlinmaxwell 29th May 2021 12:14 pm

Re: Advice on this linestage build
 
Quote:

else why use valves?
Because they are pretty, that's why I built a few valve amps, now I use a TDA7293, works a treat.

Jez1234 29th May 2021 12:49 pm

Re: Advice on this linestage build
 
FWIW I've designed and built amps in which an EF184 was giving around x 400 gain.

A CCS was not a practical suggestion with a pentode for the reasons David outlines above but I have used bootstrapping to the same effect and without the DC issues.

nzoomed 9th Sep 2021 1:53 am

Re: Advice on this linestage build
 
1 Attachment(s)
Been a while getting back sorry, but here is my updated design based on roberts post, it has a switch to bypass the baxandall tone stack and had balance controls added.

Let me know what improvements can be made.
I may consider switching out the ef86 for an ef184 if its a better option.

Radio Wrangler 9th Sep 2021 7:08 am

Re: Advice on this linestage build
 
I'm not sure what that circuit would do. Your final triode is a cathode follower configuration giving no voltage gain so the loop gain has to be below unity, and no inversion. So with positive feedback it isn't really the usual Baxandall circuit. The amount of boost is not going to be much at all.

Watch out for DC on your bypass path versus whatever the tone control network floats at. I'd expect operating the bypass switch to create a loud noise.

David

nzoomed 13th Sep 2021 5:35 am

Re: Advice on this linestage build
 
Quote:

Originally Posted by Radio Wrangler (Post 1405154)
I'm not sure what that circuit would do. Your final triode is a cathode follower configuration giving no voltage gain so the loop gain has to be below unity, and no inversion. So with positive feedback it isn't really the usual Baxandall circuit. The amount of boost is not going to be much at all.

Watch out for DC on your bypass path versus whatever the tone control network floats at. I'd expect operating the bypass switch to create a loud noise.

David

I see what you mean, when I inserted that baxandall stage into roberts schematic, it wasn't really designed for that kind of tone stack.

I dont need huge gain, as essentially this will be used as a buffer stage with minimal gain from mostly digital sources and an RIAA stage.

What can be done to prevent noise when operating the bypass switch?

Radio Wrangler 13th Sep 2021 6:40 am

Re: Advice on this linestage build
 
Take a look at the circuitry around all switches. What sets the DC voltage on the circitry for each position?.... even when that position isn't selected? Because if the path is only AC coupled and relies on being switched in circuit to have a resistive path to ground, then when switched out the DC voltage can drift through stray leakage, capacitors can charge, a little (compared to the HT voltage, but an awful lot compared to the signal) and so when you next throw the switch you get a loud 'crack'. A while back, someone used the phrase 'tone stack' which is mostly used in guitar amp circles. This is the sort of thing where the amplifier tears the cones out of speakers.

So look particularly at all amplifying device inputs. What maintains precisely the same DC bias on it when switches are thrown? Even if all positions of the switch are OK, then what about when the switch is transitioning from one contact to the next? It's easy to have a valve grid or transistor base floating for a moment, and you can get a pop.

Often you have to add high value resistors around the place to bleed off any leaked charge, or to keep networks out of use charged to the same potential they run at when switched in.

Sometimes you have to add extra AC coupling and leak resistors just to keep things tethered close to ground.

Similarly, try not to let any DC run down potentiometer tracks. This is the most popular way to make them go noisy. Do not use the sliders of pots to provide ALL the DC setting of a grid. Add a fixed resistor grid leak. Pot sliders skip out of contact as they are scraped along the track, and floating grids quickly drift at DC, and this too makes for noisy pots.

These are things not taught on electronics courses. These are things picked up by designers who have been around the block a few times and have learned from their mistakes. These are things missing from a fair few hobbyist and magazine designs, and plenty of professional designs. But doing this is 'good housekeeping' and makes for reliable equipment which just works smoothly. Get this wrong, and though your amp might sound great when working normally, but if anyone switches the input or a tone bypass etc, the front row of the audience is blown back out through the foyer.

Even with a simple home hifi setup, you don't want even small clicks and pops, or scratchy pots.

Good housekeeping, attention to detail, call it what you will, it makes a big difference to the end user.

In ye olden days (tm) valves and transistors were expensive. So hifi designs did everything possible to economise on them. Where multiple inputs needed a mixture of different gains and even RIAA shaping, the usual ploy was to have the first device in the preamp have different feedback networks switched around them. These systems were a right devil to try to make pop-free and many designers didn't even know to try. Eventually as circuitry became cheaper, separate input stages were used for the different types of input, and the outputs of these stages were switched. Feedback paths were left carefully alone.

Switched feedback could be done fairly well, but it took a lot of care. Have a look at the arrangements in the Quad 33 preamp.

David

Diabolical Artificer 15th Sep 2021 5:15 am

Re: Advice on this linestage build
 
After messing about trying to get a valve tone stage to work on one of my amp builds, I've come to the conclusion it's only an exercise for masochists. I've had great success incorporating Rod Elliots opamp based tone stage into a valve preamp and another amp, but the valve tone controls, sheesh.

The problem is you have high voltage DC going through pots or you try and block said HV with caps, but the caps add extra phase shift, which can make things unstable and implementing a bypass SW isn't trivial, as you've found, though I managed it with mine but differently to how you did it. Where you put the volume pot in the circuit effects things too, that's what I found. I used the circuit back in post one,altered a bit.

So in a nutshell I wouldn't use valves again in a tone control. A one two three four...so it's no nay never oh never no more, will I use valves in a tonestack, oh never no more ( all this sung at full whack to the tune of The Wild Rover) the experience aged me immensely : )

Andy.

wd40addict 15th Sep 2021 12:15 pm

Re: Advice on this linestage build
 
I'd have to disagree with Andy on this. Twice I've designed Baxendall tone circuitry for PA amps. One was active based around an EF86, the other passive. The important thing was that I simulated both circuits in Spice in context, with source and load impedances, and they both performed as expected once built.

Cutting and pasting a tone design into a different amplifier is not likely to be that successful.

Re the raw performance of the EF86 in pentode mode I once measured a gain of 200 with 0.1% THD. I didn't try triode strapped, but the idea (myth?) it has good performance in this mode dates back a long way. I remember a colleague who was an enthusiast back in the 50s telling me this.

Diabolical Artificer 16th Sep 2021 5:14 am

Re: Advice on this linestage build
 
Quote:

I'd have to disagree with Andy on this. Twice I've designed Baxendall tone circuitry for PA amps
Perhaps that's the difference, having a level of knowledge to design a tone control from scratch is what I lack. That and most PA/guitar amp tone stacks I've seen don't care about DC in pots, it's cheap pots, low component count, price is King all the way. Oh, and the ability to use Spice.

Andy.

Radio Wrangler 16th Sep 2021 5:59 am

Re: Advice on this linestage build
 
Spice is learnable. It's an answer to all sorts of problems rather than being a problem in itself. It is a fair step in learning it, but you don't have to absorb the whole thing in one go. You just need to do something simple and learn how to not turn on all the things you don't need. They'll wait til you have a use for them later.

The problem is that there are many half-assed simulators which look lots easier, and divert people from starting with a full-blooded one. They fairly soon hit the limitations of the 'easy' ones and get bored with them or frustrated. Some go on to look for and use a slightly more elaborate simulator and find they have to learn it from scratch.

It makes sense to curse a pox on all their houses and go learn one that will grow with you. If only it didn't look more difficult than it is....

Back to Baxandalls:

Peter Baxandall wrapped an existing sort of passive tone control around an amplifier such that its pots could swing between attenuating the input through to reducing the feedback and revving up the amplifiers gain. He'd created a circuit with fairly symmetrical capabilities. You could arrange one to be capable of as many dB of boost as it had of cut, and you could choose those limits. Rather neat, and it out-performed the passive version sufficiently that it became the dominant circuit for many years.

The passive circuit cannot give gain, so you must choose how much pseudo-boost you want with the pots screwed fully clockwise, and that is also the minimum loss you can have with it all set to flat.

Baxandall's active circuit has its limitations, too. It cannot provide more gain in the peak boost condition than the raw amplifier it uses can manage. Worse, the raw amplifier probably needs enough gain in hand at peak boost, that there is still enough feedback around it to help with its distortion and to control its output impedance. You can get a bit more gain per valve than you can get per transistor. Doing a Baxandall with the usual 12dB or so of boost doesn't look too bad with a good gainy valve, but the impedances you need to run fight you. With care, you can make something reasonable, but anyone doing distortion measurements with the pots much away from flat will help fuel the audiophile belief that tone controls are the spawn of the devil (FX: Thundercrack in the distance). Doing one with a single transistor is appreciably more difficult. Do-able it seems, but actually worse-performing.

But transistors have an ace, they are cheap and you can easily use two or more. You can have enough gain to make a good Baxandall without bad impairments at much boost, and you can avoid driving yourself down into the noise. You could have multiple valves, but the costs and heater power discourage people. Multi-transistor amplifiers are also availably in handy, easy to use, OP-amp form. You don't have to stick to the Baxandall curves, You can have parametric EQ if you want, like a zillion mixer desk channels around the world. You can even have, shhh whisper, a graphic equaliser (FX: BIG thunder crack!). Of course, that's being silly and going too far, but you can have one in your system. It doesn't need to be connected, just on power and lit up. It will then radiate its audiophile repulsion field and protect you from nuisance.

David

Diabolical Artificer 17th Sep 2021 5:47 am

Re: Advice on this linestage build
 
Quote:

Spice is learnable. It's an answer to all sorts of problems rather than being a problem in itself.
Aye David, but to be honest I can't be ars*** to learn it, prefering hands on bench testing. Manipulating physical devices gives me FB via my hands and other senses, SPICE and other PC software is all in the realm of the head, like playing chess without a board. Not for me.

My rebuke to folk who don crucifixes and bulbs of garlic when encountering a tone control on an amp is my ears are fitted with LP filters, why would I not want to compensate for this filtering? We use Daddy's & Tomato sorse on our food QED. So I fit most of my amp builds with tone controls and use opamps, those handy gain triangles.

Andy.

Radio Wrangler 17th Sep 2021 7:06 am

Re: Advice on this linestage build
 
Spice is a tool. You don't have to use it all the time and for everything, but it's valuable if you have it so you can use it when you hit appropriate problems.

The funny thing is that there are many people with no real understanding of circuits who put something into spice and muck around with values or re-connecting thing almost blindly hoping to get something to their taste. Spice is a tool for simulating things and telling you what you've got does. It's absolutely lousy at steering you which way to go if what you get isn't good. It doesn't have a function which tells you "This circuit is a piece of crap, why not try a ....."

It comes into its own when someone has a bit of a clue about what's going on.

It allows you to try ideas out without burning through your stocks of components or using up much of your life expectancy.

Like any tool, used in the right circumstances it helps, in the wrong circumstances it can make things worse.

Did you use an electric drill to make all those holes in that chassis, or did you use a hand-powered drill, and file them all larger? You've made a lot of holes in chassis in the time I've known you on the forum. On the other hand, consider a food mixer/processor. It does a two-minute job in 15 seconds, then takes 10 minutes to clean for the next job. They don't mention that part in the glossy adverts. Not all tools are good, but some are.

Spice is a useful tool. It can help your understanding of circuits, and you're at a phase where it really would help you.

David

joebog1 17th Sep 2021 7:16 am

Re: Advice on this linestage build
 
It does a two-minute job in 15 seconds, then takes 10 minutes to clean for the next job. They don't mention that part in the glossy adverts. Not all tools are good, but some are.

I LOVE the "quote" David, AND I like your humour! I have tried mucking about with spice, and I do make a VERY nice Vindaloo, :) .
I hear Andy loud and clear, and am probably in the same boat as I age. It needs a "certain" brain to work around, probably why I spent 40 years working as "hardware engineer" .
Design circuit boards I can do, transformers used be fun, design cables that dont exist I can do. ngspice ???? no can do.

VERY nice effort again Andy, and dont stress the spice.

Joe


All times are GMT. The time now is 10:24 am.

Powered by vBulletin®
Copyright ©2000 - 2022, vBulletin Solutions, Inc.
Copyright ©2002 - 2021, Paul Stenning.