UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Other Discussions > Homebrew Equipment

Notices

Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members.

Closed Thread
 
Thread Tools
Old 28th Jul 2020, 6:08 pm   #1
yestertech
Nonode
 
yestertech's Avatar
 
Join Date: Dec 2002
Location: Coulsdon, Surrey, UK.
Posts: 2,109
Default First home brew valve amp in 40+ years - disappointment

Preamble...
From early days mucking around with EL34s using tiny PP output transformers and single ended amps built with frame output transformers on the chassis of B/W teles, not to mention HT voltages which in retrospect were astonishingly dangerous, I've always enjoyed building amps with plenty of 'glass'. Class A single ended always seem to sound very 'pure' back in the day
Fast forward over 40 years and I thought I might dip a toe in the water again. This was brought about by the acquisition of 2 rather superior Parmeko 'Atlantic series' P2925 output transformers designed for P-P loads of 8K with UL taps. Before said company disappeared I managed to get a copy of their 'record card' for these beauties, which shows them to be rather excellent.
Clearly designed for something like the Mullard 5-10, but I don't 'do' EL84s (boring) so my thoughts turn to 6V6 or thereabouts, which electrically is pretty similar.

The MO...

After much faffing about ( i.e. work ! ) I finally got to knock up the test amp using an old chassis from a Vortexion tape machine. To save construction time (eager for the results ! ) this was to be hooked up to outboard power supplies for HT, LT and fixed bias for the o/p valves.
Good quality ceramic valve holders were fitted. By this stage the final valve line up had altered:
6BR7, ECF82, EBL21 x2
Unconventional, but EF86 is rather boring and I have lots of 6BR7 which is pretty similar. The ECF82 phase splitter is the 'improved' circuit designed by one A.Radford and has demonstrably superior HF performance. Instead of the predicted 6V6 output stage, I opted for EBL21, for no other reason than they look rather classy. The loading is slightly wrong ( 9k A-A rather than the 8K of the Parmekos but still I thought worth a go.

The results..

If I'm honest , quite disappointing even after making allowances. At 300V HT , The output only managed to hit around 5 watts before clipping started. The frequency response at HF began to droop north of 10kHz
The Radford design note includes step phase correction at both frequency extremes to keep the feedback from becoming positive inserted around the first stage and between it and the phase splitter. There is as yet no global feedback applied as I wanted to see how good the transformers were without. Because of this there was also too much gain so the 6BR7 stage was reconfigured as triode. Given the poor HF response on test, the correction circuit across its anode load was removed. This improved the HF a little but it was still merrily rolling off at 16kHz. The transformer spec states 20-20kHz +/- 1 dB so I would have expected something of this order prior to applying global feedback.

Whilst granted I've taken a few liberties with the circuitry, I would have expected a better showing than this.

Any thoughts as to where to go next would be welcome......


Andy
Attached Thumbnails
Click image for larger version

Name:	P2925.jpg
Views:	328
Size:	115.5 KB
ID:	212100  
Attached Files
File Type: pdf ATLANTIC P2925 record card.pdf (1.28 MB, 211 views)
File Type: docx radford phase splitter extract.docx (655.2 KB, 113 views)

Last edited by yestertech; 28th Jul 2020 at 6:23 pm. Reason: add pics
yestertech is online now  
Old 28th Jul 2020, 6:40 pm   #2
GrimJosef
Dekatron
 
Join Date: Sep 2007
Location: Oxfordshire, UK.
Posts: 4,310
Default Re: First home brew valve amp in 40+ years - disappointment

Can you measure the drive signals at the output valves' grids without loading them too much ? If they're well behaved (not clipping too soon or rolling off at HF) then you can concentrate on sorting out the output stage. A 15W valve amp driving a 15ohm resistive load will give you 15V RMS (obviously). If you centre-tap the load (you might have to float one amp or other w.r.t. earth, temporarily of course) then that should give you a nice clean drive signal to test the output stage on its own.

Cheers,

GJ
__________________
http://www.ampregen.com
GrimJosef is offline  
Old 28th Jul 2020, 7:13 pm   #3
Guest
Guest
 
Posts: n/a
Default Re: First home brew valve amp in 40+ years - disappointment

It is difficult and expensive to make a valve, 'speaker driving, audio amplifier flat to a few dBs. A piece of cake using semiconductors (my current amplifier is one of those 50W per channel chips). I have built a few valve amplifiers, the best one used output transformers costing £150 each 15 years ago, all good fun but not what you want for hifi these days (I now await the brickbats flying about).

I do like to listen to the spoken word (plays etc.) and therefore have a frequent reference as to what it should sound like, music? what should that sound like?

(I will now duck behind a large concrete wall)
 
Old 28th Jul 2020, 7:28 pm   #4
GrimJosef
Dekatron
 
Join Date: Sep 2007
Location: Oxfordshire, UK.
Posts: 4,310
Default Re: First home brew valve amp in 40+ years - disappointment

I might agree about expensive - good output transformers cost money (but Andy already has those). But difficult ? OK, if we use a chip amp then that is straightforward. But it's a different type of amp 'building' I think ...

EDIT: I think if you sat as far away from the speaker as you'd normally sit from, say, a friend who'd come round for a cuppa in your sitting room then you'd be hard-pushed to distinguish between a valve hi-fi amp and a chip amp reproducing human speech.

Cheers,

GJ
__________________
http://www.ampregen.com
GrimJosef is offline  
Old 29th Jul 2020, 7:44 am   #5
Diabolical Artificer
Dekatron
 
Diabolical Artificer's Avatar
 
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,637
Default Re: First home brew valve amp in 40+ years - disappointment

I've been mucking about designing and building valve amps for the last few years with varying results , it can be a rollercoaster, the results dire after loads of work. Looking at the frequency response graph for the latest amp I built HF rolloff started at 2600hz, 3dB at 25khz, no NFB applied, and this with decent OPT's.

If you go through the whole amp doing a frequency sweep I think you'll find the frequency response for all stages also look pretty dire without NFB applied. If the amps designed properly each added stage knocks off 3dB, so the trick is making the first stage as flat as poss. This is easy with just one valve onna bench, add a coupling cap and another stage, it starts getting worse fast.

OPT's influence the amps sound and stability to such a great extent I don't think you can take an existing design and plonk a different OPT on the OP without quite a bit of testing and tweaking. First get pre OP stages with as decent a frequency response as poss, this involves a lot of maths &/or subbing caps and testing.

My advice as stated get the first stage as flat as poss then select RC's for best F response, direct couple if poss. For PP amps I've found that it's a good idea to make your phase splitter as beefy as poss = lots of V swing and plenty of current, which means an LTP using something like a 12BH7 or even two EL84's, seems mad but it works. Whichever you use if "driven" from a CCS you have a simple PS with few components.

Ignore distortion of individual stages, a bit of distortion is good, which would be blasphemy to Mr Radford, thing is if you want an amp with 0.0005% THD you wouldn't use valves, it's the added harmonics that give valve amps a nice sound and compensates for low efficiency and expense of components.

Some pots thrown into the design helps, so you can adjust, AC/DC balance going into the OP stage. A LTP with CCS is extremely well balanced, but it's a good idea to add a pot to the anode R's so you can tweak it. Add a pot to one side of a cathode biased PP OP stage.

More controversial advice, add tone controls, we're not after transparent hifi here we're after smile inducing warm fuzzy stereo. When all this is done you have to get the d*mn thing stable, which means tweaking the design again, so more testing but at frequencies on the extremes of response, so under 100hz over 20khz. Good fun.

Andy.
__________________
Curiosity hasn't killed this cat...so far.
Diabolical Artificer is offline  
Old 29th Jul 2020, 7:53 am   #6
Guest
Guest
 
Posts: n/a
Default Re: First home brew valve amp in 40+ years - disappointment

Oh yes, add tone controls.
 
Old 29th Jul 2020, 9:00 am   #7
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: First home brew valve amp in 40+ years - disappointment

I've done valve-based amplifier stages to well over 100MHz, so why should stages roll off before 20kHz? This is fixable if you want to fix it.

I agree that in normal designs successive stages roll off quite a bit and that the resulting amplifiers rely quite a lot on overall feedback to tart-up their frequency response to something reasonably flat.

The limitation on the bandwidth of the small signal stages lies in trying to get lots of gain out of each stage by running devices in high impedance environments. Because the maximum Gm available from valves is a bit limited, reducing operating impedances brings stage gains down a lot, so you have to compensate with more stages. Local degeneration (cathode resistors split with any decoupler part way down) improves linearity and bandwidth, but lowers effective Gm. The improvement in linearity removes some of the 'valve sound' of course. Once you get into wider stage bandwidths and using balanced push-pull from earlier in the chain, your amplifier starts to look much more like an ouvre by Tektronix than by Leak.

With what you've got, Use an oscilloscope with a pair of high-impedance x10 probes and take a look down your open-loop amplifier. At the output of each stage play about with the test signal frequency and amplitude. Find where it rolls off at high and low frequencies, take it up to find the clipping point. Are the phase splitter halves running roughly equal anode currents?

You have to go actively looking for trouble to find it in these circumstances, it doesn't stick its head up and say hello. Play around and look all over the amplifier until you find something that doesn't seem right.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 29th Jul 2020, 2:13 pm   #8
Diabolical Artificer
Dekatron
 
Diabolical Artificer's Avatar
 
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,637
Default Re: First home brew valve amp in 40+ years - disappointment

"The limitation on the bandwidth of the small signal stages lies in trying to get lots of gain out " That's the problem, in a scope you only have to boost the signal to a few dB, if that, but in an AF amp we often have a gain of 1000 or so before the OPT. The layout of a AF amp often doesn't help, distances between valves mean long leads and stray C, unlike most valve scopes where the distance between stages is an inch or so. The OPT puts a big kibosh on things cos with it in the closed loop we have to stomp on the LF and HF gain to stop the amp singing. An AF amp with an OPT is like driving a bike, throttle wide open with the brakes on.

Andy.
__________________
Curiosity hasn't killed this cat...so far.
Diabolical Artificer is offline  
Old 29th Jul 2020, 3:05 pm   #9
Skywave
Rest in Peace
 
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
Arrow Re: First home brew valve amp in 40+ years - disappointment

Quote:
Originally Posted by Radio Wrangler View Post
With what you've got, use an oscilloscope with a pair of high-impedance x10 probes and take a look at your open-loop amplifier.
Play around and look all over the amplifier until you find something that doesn't seem right.
I'd like to expand on that above quote.
Although it's been many years since I trod the path of developing / building AF amps., I was always taught that it is essential to get the open-loop characteristics of an amplifier correct first, prior to adding (negative) feedback. In summary, the overall idea of N.F.B. is to improve the desirable features of an amp. whose features are 'good' in the first place, not to correct short-comings in the basic, original design.
Hence, when investigating causes of poor performance of such an amp., do so without any N.F.B. that is operational. The existence of that N.F.B. will probably mask the deficiencies that you are then looking for.

Al.
Skywave is offline  
Old 30th Jul 2020, 9:15 am   #10
cathoderay57
Nonode
 
Join Date: Oct 2017
Location: Bristol, UK.
Posts: 2,364
Default Re: First home brew valve amp in 40+ years - disappointment

I stopped worrying too much about flat responses at the HF end when I discovered that age-induced hearing loss had curtailed my ability to hear anything above 10kHz. Just a thought! Cheers, Jerry
cathoderay57 is offline  
Old 30th Jul 2020, 11:06 am   #11
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: First home brew valve amp in 40+ years - disappointment

Andy,

The CRT in the scope on my bench here has a Y sensitivity around 8 volts/centimetre.
On the the Y-scale knobs I can select 0.005V/cm so that's a gain of 1600 or 64dB with 100MHz bandwidth. That's the gain of the amplifier. All other settings come from an attenuator in front of the amp.

In the days of valve scopes, bandwidths were not usually so high. 20 or 50MHz was considered good in a lab scope, but without the PDA tubes with domed mesh electron lenses, or multi-section distributed deflection plates, the CRT sensitivity was a lot lower, so the overall gain had to be higher though the bandwidth was lower.

On my trusty 1740, there's a x5 button which switches the amplifier gain up, at the loss of some bandwidth, and that gives 1mV/cm so the gain is around 8000 = 78dB

So scopes need high gain amplifiers and they need to be flat open loop, with good pulse responses etc etc.

The limitation in audio amplifiers lies in the gain per stage they are trying to get. They limit themselves in the number of stages they're prepared to use.

Pushed for the highest gain in a stage, that stage will suffer reduced bandwidth and increased phase shift. Replace that stage with two stages of lower gain and you'll get better bandwidth and phase shift, but then you get the problems of two stages multiplied together. However, it happens that the combined impairments of the two stages is still usefully better than those of the one-stage attempt.

Valve audio amplifier design, with a few exceptions is trapped in a timewarp of input stage, phase splitter, output stages. It was financially necessary back when valves were very expensive and were thought to have rather limited lives.

Harold Leak worked hard to get this structure down to about 1% distortion open loop with a transformer good enough to allow enough feedback to bring the closed-loop performance down to his target of 0.1% distortion. Looking back through measuring some of these things, he only just made it. He was having to juggle gain, bandwidth, stability margin and distortion all at once.

You can break out of this three-stage straitjacket if you want. If your parts came out of your junkbox, price isn't important. Your amplifier will lack in apparent simplicity, and the appearance of simplicity has long been seen as a virtue. You'll still be limited by your output transformer in terms of how much feedback you can apply, but now the valve stages aren't chucking in extra phase shift and bandwidth limitation. You can go even further, ditching the transformer and using a ludicrous number of high-current valves. The whole thing becomes monumentally inefficient, and about as big as, well, a monument - maybe a small war-memorial.

Once you've done all this, you've designed-out all the 'valve sound' except for the expectation in the head. Your brain sees valves and logic says that whatever you hear has to be valve sound. It all gets messily psychological from here onwards.

As Skywave says, there is no substitute for getting the open loop performance good. The amount of improvement you can summon by way of feedback is limited, so if you want a good end result, you can calculate where you need to start from.

Those gorgeous valved scope from Tektronix had no overall feedback. Modern transistor ones don't either.

But wait a minute... aren't transistors supposed to be totally evil distortion generators?
Well, they certainly have abrupt non-linearities in their characteristics, and something is needed to fix this. The answer is local degeneration, usually by undecoupled emitter resistors. Look in that Tek scope and you see undecoupled cathode resistor networks.

To do a valve amplifier right to the edge of the performance possible with a small number of stages needs you to really know what you're doing to a Terman/Bailey/Baxandall sort of level, and it's still going to take development work.

The answer is to cheat. Take away the limitation on the number of stages and you can get that same performance by living less dangerously. Demands on your brain are less, and the stability margin is more reliable. But this is hifi, people accept no limits. no amount is ever enough so here comes the pressure to push the new architecture to its limits. It is a battle without an end.

It did dawn on me that once I'd designed out the 'valve sound', I might just as well have used transistors and it would have made the output transformer very easy to dodge.

MM - tone controls are great. As well as providing a useful function in allowing you to adjust things to your taste without having to replace equipment, they also stop audiophools dead in their tracks and throw them off the trail.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 30th Jul 2020, 11:25 am   #12
stevehertz
Dekatron
 
stevehertz's Avatar
 
Join Date: Sep 2011
Location: Rugeley, Staffordshire, UK.
Posts: 8,809
Default Re: First home brew valve amp in 40+ years - disappointment

Just a couple of random points. Age related hearing loss means - for me at any rate - that anything above 10-12k is pretty much a waste of time. I agree totally about the spoken voice being the best test for accuracy. Music is music, it is comprised of instruments whereby if their tonal characteristics are changed even quite a bit, it doesn't really harm the music or maybe not even register a change to us. Conversely, we hear the natural, un-amplified human voice each and every day and we can easily differentiate a real one from an electrified one.
__________________
A digital radio is the latest thing, but a vintage wireless is forever..
stevehertz is offline  
Old 30th Jul 2020, 12:17 pm   #13
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: First home brew valve amp in 40+ years - disappointment

Michael Flanders summed it up nicely:

"Still, one ought to please any passing bat,"

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 30th Jul 2020, 12:18 pm   #14
Skywave
Rest in Peace
 
Join Date: Jun 2006
Location: Chard, South Somerset, UK.
Posts: 7,457
Default Re: First home brew valve amp in 40+ years - disappointment

My understanding of the problems encountered here hinges on a simple but fundamental concept: gain-bandwidth product. For a given configuration, if one factor of that product is increased, the other must obviously decrease. Any amplifier will consist of a number of amplifying stages. For example, if the overall bandwidth is inadequate and needs to be increased, then the overall gain must be reduced - and vice-versa. However, an increase in the number of amplifying stages will increase the overall gain-bandwidth product, thus giving the designer more 'space' to trade-off one variable against the other.

Al.
Skywave is offline  
Old 30th Jul 2020, 12:21 pm   #15
PJL
Dekatron
 
Join Date: Sep 2005
Location: Seaford, East Sussex, UK.
Posts: 5,997
Default Re: First home brew valve amp in 40+ years - disappointment

The benefit of the Radford LTP is the EF86 is driving a pentode so there is no Miller capacitance to deal with. Do we get to see the full circuit? EBL21's seem an odd choice for some high quality Parmeko transformers with UL taps.
PJL is offline  
Old 30th Jul 2020, 1:15 pm   #16
yestertech
Nonode
 
yestertech's Avatar
 
Join Date: Dec 2002
Location: Coulsdon, Surrey, UK.
Posts: 2,109
Default Re: First home brew valve amp in 40+ years - disappointment

My sincere thanks to everyone who has contributed to this thread so far – I’m grateful for the very considered replies and will be back on the bench as soon as time permits to take some measurements on the existing circuitry.
I do have to concur with the view point that the open loop gain and other characteristics without global feedback should be reasonably good. At least, that’s what I was always taught many moons ago. This was the main reason why I knocked up this particular amplifier as in the past I have always been constrained by cheap and nasty output transformers, Whereas this time I have some top quality ones with a known excellent performance.
I can post the full circuit but it’s more or less what is in the original post with a couple of fixed bias output Valves fed by the output of the phase splitter.
I would entirely agree about the odd choice of EBL21: they were picked for no other reason than as a metal-based straight sided Valve they look nicely retro, akin to the original metal-based EL 34s. Not exactly a good technical reason! I wanted something which wasn’t EL84 /6V6 but was roughly in the a-a loading range of the transformers. Their spec states that they should deliver 13 watts at 1.7 %. So my hope was that in ultralinear with a small mismatch in the loading I might just achieve 10 at lower distortion to which I could apply some global feedback.

I will get back to the bench and some measuring as soon as I can and hopefully make any necessary adjustments to get a half-decent amp

Thank you again to everyone who has taken the time to contribute to this thread

Andy



.

Last edited by yestertech; 30th Jul 2020 at 1:20 pm.
yestertech is online now  
Old 30th Jul 2020, 2:09 pm   #17
PJL
Dekatron
 
Join Date: Sep 2005
Location: Seaford, East Sussex, UK.
Posts: 5,997
Default Re: First home brew valve amp in 40+ years - disappointment

Have you included grid stoppers?
PJL is offline  
Old 2nd Aug 2020, 3:19 pm   #18
yestertech
Nonode
 
yestertech's Avatar
 
Join Date: Dec 2002
Location: Coulsdon, Surrey, UK.
Posts: 2,109
Default Re: First home brew valve amp in 40+ years - disappointment

Not at this stage, no. No stoppers in the a or g2 connections either.

Andy
yestertech is online now  
Old 2nd Aug 2020, 3:29 pm   #19
Radio Wrangler
Moderator
 
Radio Wrangler's Avatar
 
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,800
Default Re: First home brew valve amp in 40+ years - disappointment

If you do get parasitic oscillation at frequencies above audio, the only clue you may get could be off behaviour at DC or distortion without any logical explanation.

Stoppers are considered good housekeeping and normally included without needing evidence that they are definitely needed. It's like skating on a lake. If you finish the day alive, the ice was thick enough... but you don't know what the safety margin was.

David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done
Radio Wrangler is offline  
Old 5th Aug 2020, 11:51 pm   #20
joebog1
Banned
 
Join Date: Jun 2015
Location: Mareeba, North Queensland, Australia
Posts: 2,704
Default

I have had similar problems!!. I found quite a few articles by Joseph Marshall in various American periodicals. I have studied them all, and believe him to have been a bit of a master at designing.
Here is one of his articles that, unfortunately seconds everyone else's suggestions in this thread.
But it will enhance your knowledge anyway.

Joe
Attached Files
File Type: pdf Extending amplifier bandwidth1.pdf (368.7 KB, 76 views)
File Type: pdf Extending amplifier bandwidth2.pdf (295.1 KB, 64 views)
File Type: pdf Extending amplifier bandwidth3.pdf (209.1 KB, 65 views)
File Type: pdf Extending amplifier bandwidth4.pdf (485.2 KB, 77 views)
File Type: pdf Extending amplifier bandwidth5.pdf (486.0 KB, 74 views)

Last edited by Station X; 6th Aug 2020 at 9:37 am. Reason: Links changed at poster's request.
joebog1 is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 12:50 am.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.