Re: THD Univalve Mods and Rebuilds

[uneumann]

The univalve is an older amp that makes a great mod/rebuild platform (IMO). I'm a fan of the class-A sound. I've done some work with results I'm really happy with. The saga unfolds at https://sites.google.com/site/stringsandfrets/Home/univalve-mod - with lots of pics and sound clips. Anyone out there have an THD stories with happy (hopefully) endings?

[pdf64]

Well done - interesting stuff.
Re the turnover frequency of cathode RC, my understanding is that the tube's internal cathode impedance should be taken into account. It's effectively in parallel with the cathode resistor and is = 1/gm. So when you bring it into play, it will move up the turnover frequency of the cathode circuits.
See
Designing Common-Cathode Amplifiers
especially appendix B.
Pete.

[uneumann]

Well, my calculations did not take that effect into account. I can imagine (as mentioned at the bottom of the link) that any additional capacitance in parallel with the cathode resistor would reduce (not increase) the roll-off freq - right? Either way, I didn't take it into account. That capacitance is pretty small, right? So it's impact is negligible at audio frequencies.

[pdf64]

1/gm = ~ 625 ohms, so it makes a significant impact on the turnover frequency. Pete.

[uneumann]

Hi Pete - are you talking about output impedance?
As I understand it, that's a separate issue from how the cathode bypass roll off is computed - but I can be taught new things... please show me an example.

How do you compute the 3db point given a common case: a 12ax7 with 1.5K cathode resister and a 2.2uf bypass?
What calculation would you do - and what roll off freq comes of that case?

[pdf64]

1500 // 625 = 441 ohms
1/(2 x Pi x 441 x 2.2e-6) = 164 Hz
Have a good read of that Aiken link. Pete.

[uneumann]

Hi Pete - thank you. I have neurons that knew these models many years ago, but they have atrophied since this is no longer my main livelihood. They thank you for resurrecting them and making them productive members of society again. Reading the Aiken link brought back textbook memories. It appears that I am in for a round of recalculation and probably some economic stimulus if I decide to procure some new parts as a result of the new calculations.

I believe simpler models hold for transistor circuits in the audio frequencies. I have used near-ideal models of voltage sources (emitter) and current sources (collector) for many year now with pretty good success. Not so for tubes, I now know. I also understand that high-frequencies (>100KHz - the black arts) require better models for both transistors and tubes.

Thanks again...

[pdf64]

Yes, when I've not concerned myself electronics projects for a year or 2, I've been surprised what a struggle it is to get up to speed again. Every year I annoy my brother when I ask him the garden plants are as they come into flower, there's not much chance that I'll have remembered it in 12 months time.
Re higher frequencies, it's amazing what an significant effect the grid to plate Miller capacitence, combined with grid circuit impedance, has >1kHz; makes me appreciate the difficulty of designing for hifi - certainly no 1M vol pots there. Pete.

[Steve Conner]

Appendix C: The math behind the lower cutoff frequency due to a partially-bypassed cathode

This section under construction - thanks to Jean-Pierre Trolet for pointing out an error in the original document

This is somewhat controversial. I always assumed it was a shelving response, and as such it has two cutoff frequencies. The upper one is determined by the cathode resistor in parallel with the tube's internal cathode resistance, as Pete says, the lower one is determined by the cathode resistor alone as suggested in the original post.

I have used this method with transistor circuits, but Merlin B. pointed out to me that it doesn't work with tubes. With the values used in tube circuits, the two cutoff frequencies are so close together that the assumption of two cutoff frequencies isn't even valid any more.

Put another way, RC time constants need to be separated by at least a decade in frequency to be analyzed independently. But in tube amps you rarely get a shelving boost of more than 6dB, or one octave separation. In this condition, Pete/Merlin's method gives a more sensible answer.

[uneumann]

OK - I think I have this figured out... I had posted a response earlier and it was not right so I deleted it and here is what I've found and I *currently* believe to be true. :-)

There are indeed two break points in the CC response due to a cathode bypass cap. This creates the shelve response shown on Aiken's site and elsewhere. The two points of interest are the LF point of the upward response inflection, and the HF point where the curve flattens and becomes level. There is also a point of interest in the center - the mid point of the transition zone - which I'm not going to pursue here but as a matter of interest someone else has figured that out (although I've not verified it).
See http://www.freewebs.com/valvewizard2...BypassCaps.pdf if you're interested.

Back to the two breakpoints... Steve, you're correct in that I've computed the LF point, and my mistake was in thinking it was the HF point.
Pete's pointer (and Aiken's site) describes computing the HF point, which is the point of interest to me. Pete, your estimate of the cathode impedance is 1/gm ~= 625 ohms, which according to Aiken and the site below is approximate, since it doesn't account for Rp. He points out that using Rp gives a better estimate of (ra+Rp)/(mu+1) ~= 1609 ohms for a 100K Rp case. I like that better... :-) although it still impacts my numbers significantly.

This still leaves the question you raise, Steve, about Merlin B suggesting this doesn't work with tubes...
I found a counter example and seemingly knowledgeable source at Poles, Zeros, and Cathode Bypass Caps | ken-gilbert.com. Near the bottom he makes a clear case that a) this is a correct interpretation and model, and b) it is perfectly correct for tubes, and c) tube circuits need not have these two points near each other. The issue of gain comes into play since a 6db/octave slope between the two points may not be realized by an actual device since the upper point can become gain limited (effectively moved left).

I'm going with this - it seems like a good analysis to me and it makes sense. I hope I interpreted and paraphrased your comments correctly.

[uneumann]

A brief update... A new Univalve mod (ModD) is posted. In addition to some circuit improvements, the bottom of the page has scope waveform pictures at key stages of the amp. If nothing else you may enjoy seeing what happens to the signal as it flows through the amp... :-) See what you hear.

https://sites.google.com/site/string...alve-mod/mod-d

[ficelles]

I know this is an old thread but I wondered how many people had tried uneumann's ModA and used the amp for bass? I'm after an SVT sound with a 6550 or KT88... I still want to use the amp for 6 string as well of course so don't want to change the top end too much

ficelles

[ficelles]

I know this is an old thread but I wondered how many people had tried uneumann's ModA and used the amp for bass? I'm after an SVT sound with a 6550 or KT88... I still want to use the amp for 6 string as well of course so don't want to change the top end too much

I'll answer my own question here - I've implemented most of Mod A and the results are outstanding both for guitar and bass.

Specifically I first removed the bypass cap C7 which resulted in a much mellower sound for guitar but one that you can dial higher frequencies and edginess into at will using the treble and attitude controls.

I then replaced C3 with a 10uf cap and tried it with bass through a 1x15 Ampeg cab - the result is just what I'd hoped for so no need to shell out for a B15. You also get a much bigger low end with 6-string through a guitar cab.

So two really simple mods which not only improve the amp if you want more low end but are also easily reversible.

ficelles