|
Homebrew Equipment A place to show, design and discuss the weird and wonderful electronic creations from the hands of individual members. |
|
Thread Tools |
21st Jun 2019, 6:54 am | #61 | |
Nonode
Join Date: Jun 2016
Location: Bognor Regis, West Sussex, UK.
Posts: 2,294
|
Re: Valve preamp.
Quote:
I simulated a version of Fred Nachbaur's pre-amp, modified as I described above to use a negative supply, on LtSpice. I attach the Schematic and frequency plot. The Schematic has dc voltages on the principle nodes. You will see that the long tailed pair is balanced and the amp seems stable. The voltage source shown on the schematic (where Fred has neons) was just for convenience of modelling, if the pre-amp was ever to be built this should be replaced by two neons in series or a 150V zener. Interesting Peter |
|
21st Jun 2019, 7:49 am | #62 |
Nonode
Join Date: Jun 2016
Location: Bognor Regis, West Sussex, UK.
Posts: 2,294
|
Re: Valve preamp.
I should add that I used ECC83 (12AX7) for all positions as I did not have a spice model for the ECC81 handy.
I will also note that I only modelled the neons/zeners in the dc feed back loop to keep it as close as possible to Fred's original design. If I ever build it I will drop these and use straight forward resistive division, ie a single reisistor of 560K from the anode of the last valve to the junction of R111/R110 (Fred's numbers) and R110 is 270K to -120V One of these resistors is then trimmed to set the voltage balance on the diffential pair. (Could make one of them a resistor + pot) The only reason for the neons in the first place is because in the original design R110 goes to 0V and the resulting division to get down to the 27V of the grid of the differential amps would reduce the gain too much for effective dc control. Use of the -120V for the division removes the need for the neons. Peter |
21st Jun 2019, 1:12 pm | #63 |
Dekatron
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,657
|
Re: Valve preamp.
Thanks again Peter, will give it another go, probably being dense here but re the response graph which is the response trace, the darker upper one? In general terms what did you think of Fred's design? I attached both schematic and graph for easy quick glance. Fred Nachbaur died a few years ago unfortunately but some his work and design's are available here for those interested - http://www.dogstar.dantimax.dk/tubestuf/index.htm
Andy.
__________________
Curiosity hasn't killed this cat...so far. |
21st Jun 2019, 5:21 pm | #64 |
Dekatron
Join Date: Sep 2005
Location: Seaford, East Sussex, UK.
Posts: 5,997
|
Re: Valve preamp.
If it's clipping at such a low output something is wrong. Check the valve voltages.
|
21st Jun 2019, 6:37 pm | #65 | |
Nonode
Join Date: Jun 2016
Location: Bognor Regis, West Sussex, UK.
Posts: 2,294
|
Re: Valve preamp.
Quote:
I am busy this weekend but will answer your question as to what I think of the design monday. Peter |
|
24th Jun 2019, 1:15 pm | #66 |
Nonode
Join Date: Jun 2016
Location: Bognor Regis, West Sussex, UK.
Posts: 2,294
|
Re: Valve preamp.
Hi Andy,
Further to my previous post and in answer to your question as to what I think of Fred's design. The basic principles are sound, the use of an operational amplifier with the RIAA filter in the feedback is standard practice. The use of valves to form an operational amplifier using a differential amplifier input followed by a common cathode amplifier is fine. Where I have issues is with the compromises needed to cope with a lack of a negative supply. I set out the issues in a previous post but in summary: 1 Poor common mode rejection and the use of un-balanced anode resistors in the differential amplifier 2 Having to have the grids of the differential amp at +27V and the dc blocking capacitors needed in consequence. 3 the need to use neons as voltage dropping elements for the dc feedback loop. Restoring the negative supply and removing the compromises makes it much better circuit. I attach a revised schematic without the neon / zener element in the feedback. If you do build it and find the anode voltages of the differential amplifier are un-balanced adjust the value of either the 560K (R4 in my Schematic) or the 270K (R2 in my schematic) resistors. Or change the 560K to a resistor plus pot and tweak for balance. I also suggest adding the final 47K resistor and 100pF capacitor (R116 and C108 in the original circuit) to ensure HF stability. Let me know how you get on if you build it. Peter |
25th Jun 2019, 1:48 pm | #67 | |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,077
|
Re: Valve preamp.
Quote:
Many circuits exist, and have the mid-band gain fairly high so that by the time the inevitable HF zero occurs, the frequency is out of the AF range. But it's still there. As for neons as level-shifters... Aargh! Though it is easier and cheaper than an additional negative supply. But it does make the whole thing rather sensitive to supply-rail variations, whereas two rails (which generally vary in the same ratio as mains drifts up and down) are less problematic. Agree - PNP valves would be SOOOO useful. Even reverse-polarity thermionic diodes - so you didn't have to float or decouple the heater supply if you need to put RF on the cathode - would have been great. Sigh... |
|
26th Jun 2019, 4:34 am | #68 | |
Nonode
Join Date: Jan 2009
Location: Papamoa Beach, Bay of Plenty, New Zealand
Posts: 2,944
|
Re: Valve preamp.
Quote:
Here is the Quad 33 example as redrawn by Baxandall, with the supplementary pole highlighted: (I don't have a corresponding valve example on hand.) National Semiconductor did cover the issue in its LM833 audio op-amp application note as something to be addressed when very accurate adherence to the RIAA curve was required. It does not seem to have been mentioned though in connection with the earlier LM381 low-noise preamplifier IC, which could be described as having been a quasi-op amp. As an aside, but perhaps not irrelevant, although, as would be expected, the LM381 had a differential input, provision was made for single-ended operation of the input stage where lowest noise was required. By the time op-amps such as the NE5534 and LM833 arrived, they had low enough noise (at or below the level of MM cartridge inherent noise) that such artifices were not needed. But I wonder if there is some relevance to valve-based preamplifiers, where noise might be more of a problem. I should think that even when the error from not doing so is negligible, it would have been preferable to continue with the indefinite 6 dB/8ve downslope, in part as a filter for any extraneous ultrasonic signals that might come from the record/cartridge combination. The supplementary pole is easily added with one R and one C per channel, whether to a valve or a solid-state circuit. (That said, there was one application where the flattening out of the equalization curve at 20 kHz give or take that came with the series-feedback preamplifier seems to have been required. That was in CD-4 processors that used the “full-service” ICs, such as the CD4-392 and QSI-5022. These could handle the subcarrier at the very low post-equalization level, but presumably with reasonably flat frequency response in the 20 to 45 kHz range, and not a 6dB/8ve downslope across that range. The QSI-5022 had an internal series-feedback preamplifier stage, so all of the equalization had to be done there. With the CD4-392, it was customary to use an external TA7122 preamplifier IC with full equalization, or two of then with split equalization with the subcarrier extracted after the first. So what appears to have been a shortcoming in the series-feedback equalization circuit was actually turned to advantage.) Cheers, |
|
26th Jun 2019, 6:33 am | #69 |
Dekatron
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,657
|
Re: Valve preamp.
Thanks again Peter, I'll rebuild it with your tweaks after I've done with this Shure M65. I can configure the PSU to have a negaitve rail as the HT secondary has a CT. Have slowed down a bit on this project but did build the Shure M65 circuit over the weekend, but found it had too much gain on the top end, swapped C9 470p for a 1n which gave better results. Overall though response isn't flat and this circuit is the less accurate of the ones I've tested.
Which brings me neatly to ask in response to this and other posts above - " How accurate does an RIAA stage need to be?" On some schematics and builds caps like 99.3n are required which means hours of endless fun measureing caps to 3 decimal places and paralleling em or fitting big air adjustable caps and spending more hours of endless fun doing response graph's. Which leads me to another question that I've not found an adequate answer to - FB or no FB? I think David did try and answer this but I didn't understand his response. So far my tests show no FB design's to be less troublesome and more accurate. Andy.
__________________
Curiosity hasn't killed this cat...so far. |
27th Jun 2019, 1:08 am | #70 |
Banned
Join Date: Jun 2015
Location: Mareeba, North Queensland, Australia
Posts: 2,704
|
Re: Valve preamp.
As usual Andy, feedback will reduce any distortion to some degree.
My latest phono stage (on backburner till my cataracts are gone) will be the first to use feedback. I have always gone with passive RIAA until now, so it's new for me. The lashup on breadboard I did worked superbly, so I'm hoping the finished article in a my 2RU cabinet will work. Valves are on their side to get em all into the cabinet with both channels side by side. Joe |
27th Jun 2019, 7:34 am | #71 | |
Nonode
Join Date: Jun 2016
Location: Bognor Regis, West Sussex, UK.
Posts: 2,294
|
Re: Valve preamp.
Quote:
"It was suggested by Wright in The Tube Preamp Cookbook (1995) that Neumann cutting lathes incorporated this bandwidth defining filter by means of an extra, high-frequency pole at 3.18µS (50 kHz) in the cutting equaliser." http://www.pspatialaudio.com/neumann_pole.htm It is debateable however whether such a pole exists. As to which is better, passive or FB equalisation, you pays your money and you takes your choice. A quick google search will bring up arguments in favour of both but personally, and it is a personal choice, I prefer passive, no feedback circuits when using valves and tend to use feedback when using op-amps. One advantage of passive, no feedback circuits is that the circuit breaks down into sections, easily testable independantly. To answer your question about how accurate an RIAA equalisation amplifier has to be I believe that it should be as accurate as you can reasonably make it. The key word being reasonable. As you can see from the photos I posted I am prepared to parallel capacitors to get close to the theoretical values but I use 1% resistors and 2.5% polystyrene capacitors and consider that good enough. In practical terms if you end up within 1db that will be good enough unless you have golden ears (climbs into foxhole and awaits incoming fire). Peter |
|
27th Jun 2019, 9:18 am | #72 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,858
|
Re: Valve preamp.
For a cutterhead to be moved with no limitation on its bandwidth is going to take infinite power.
Any closed-loop cutter is also going to need its loop to be stable and that means having controlled time-constants within the loop, leading to a controlled lowpass function as the closed-loop response. A high frequency pole in the cutter is inevitable, the choice reduces to where the first one should be. 50kHz seems like planting it high enough that the effects in the audible range ought to be negligible. Also, a pole in the cutting process is not quite equivalent to a pole in the playing process. There is noise ingress and distortion in-between the two processes. Again, with a pole off of the top of the audible range, it should be questioned whether the change in noise can be discerned. Once you accept the prime credo of the audiophile, that there is no limit to the discernment of their hearing, then nothing can be considered negligible. Whether they can hear it or not is inconsequential. They are steered entirely by the belief that they can hear everything and this steers the flow of money from them. Accepting that transducers made of meat have limitations is aliberating thing, you can employ components with non-zero tolerances, you don't need to follow theoretical curves with absolute precision. You can ask 'Does it really matter?' Audiophiles are certain that engineers are cloth-eared Engineers are certain that audiophiles are deluded I can understand how, in the conflict, audiophiles feel more disparaged than engineers do. I'm happy to be an engineer. Equations make great armour. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
27th Jun 2019, 10:00 am | #73 | |
Dekatron
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 7,077
|
Re: Valve preamp.
Quote:
For a record cutter to allow a pole to exist, whether electronic or mechanical, because he thinks that the replay equipment may not be quite right, is wrong. It would mean that an amplifier designer who DOES get it 100% right, the overall chain would be compromised! As for how accurate does the RIAA replay curve have to be respected, well, it depends on your overall aims and the resulting error budget. If you want it all to be correct to +/- 0.5db over the entire AF range and your pickup cartridge has a level response within 0.1db (stylus velocity to output voltage), your speaker/crossover is level to within 0.22db, then you can allow your amplifier to have a maximum error of 0.18db (so using R's and C's of 2% tolerance virtually assures this). If you have speakers which show a 1db variation, then you might be dead before you start. However, if you know what the speaker variation is, you can compensate for it in your amplifier's equalisation just as the record manufacturer compensates for their cutting equipment in the recording equalisation. |
|
28th Jun 2019, 4:07 pm | #74 |
Heptode
Join Date: Jan 2006
Location: Todmorden, West Yorkshire, UK.
Posts: 870
|
Re: Valve preamp.
If you read pspatialaudio's web page a bit further you will find that they examined a couple of Neumann's disc cutting equaliser circuits and could find no trace of a 3.18uS pole.
|
28th Jun 2019, 4:49 pm | #75 |
Dekatron
Join Date: Sep 2005
Location: Seaford, East Sussex, UK.
Posts: 5,997
|
Re: Valve preamp.
Mechanical limitations of the cutter? Cutter transducer inductance?
Not sure that absolute accuracy is much of a concern as the engineer would have twiddled with the tonal balance to get their preferred result and a 'twiddle' is probably not that accurate! |
28th Jun 2019, 6:13 pm | #76 |
Nonode
Join Date: Jun 2016
Location: Bognor Regis, West Sussex, UK.
Posts: 2,294
|
Re: Valve preamp.
|
28th Jun 2019, 7:28 pm | #77 | |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,858
|
Re: Valve preamp.
Quote:
IF the maths is done, a fully accurate equalisation every bit as accurate as can be achieved with a passive network is the result. Some people baulk at the maths but construction is easy enough and you get benefits in noise and dynamic range. A 2 or 3 stage transistor job is gainy enough to use simply calculated values. Those who are limited to a single valve on religious grounds have problems anyway. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
|
29th Jun 2019, 8:11 am | #78 |
Dekatron
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,657
|
Re: Valve preamp.
I've read a RIAA stage using FB will stay accurate as the valve ages, I've found passive circuits work best in the few circuits I've tried so far, they're more stable for a start and less frolicksome.
Andy.
__________________
Curiosity hasn't killed this cat...so far. |
16th Nov 2019, 9:38 am | #79 |
Dekatron
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,657
|
Re: Valve preamp.
Made some progress on this, finally decided to make a separate line preamp and RIAA stage. The line pre will power the RIAA on a separate chassis via an umbilical cord.
The circuit is an odd one, namely a CCDA - constant current draw amplifier, the idea being from what I gather is that if an amp stage's current draw is constant, the PSU specs need be less critical, but the main reason I went with it was it's simple, has the right gain - about 10 and works. I used two VR150's to clamp the HT in order to give the 820u 400v cap an easy ride and they look pretty too as well as filling out an otherwise empty chassis. Siting and drilling holes for the 12 100k presets was a nightmare, I just don't have the tools or skill to work to high tolerances of 1mm or so. I got it wrong quite a few times so there's a few holes to fill in. Wiring in the pics is temporary as is the pot, just wanted to get everything working and take some voltage readings. Andy.
__________________
Curiosity hasn't killed this cat...so far. |
10th Dec 2019, 4:50 pm | #80 |
Dekatron
Join Date: Jul 2009
Location: Sleaford, Lincs. UK.
Posts: 7,657
|
Re: Valve preamp.
Finished bar a few finishing touches, IE labels. I tried Leraset on brass plate, letters ok but cocked up using thinned nail varnish as protection, will use laminate next time. Have to find better knobs too and some screws.
TFL, Andy.
__________________
Curiosity hasn't killed this cat...so far. |