High-ohm grid resistor in tube (pre)amp design
in [Design]
|
From: ChrisQ on 12 Nov 2009 13:45 Phil Allison wrote: >> and there's an added input to the second stage via c51 from next2, which >> comes from the first page. > > ** For the "presence" pot - as used in Marshalls as well. > > Just in case that was too much, just redrew the circuit with v4 the right way up and the presence pot included. What we seem to have is a 0.047 between the grid across r53 and r51, with the junction of r51, r57 and c31 connected to the series pot and a couple of caps, one of which is 0.033. Starts to look a bit wein bridgy to me, in a rough sort of way. Result would be a frequency dependent amplitude into V4a grid, which might produce some interesting non linear effects in the op stage. Anyway, stop being such a misery Phil and tell us how you think it works :-) Regards,
From: Richard Rasker on 12 Nov 2009 13:46 Phil Allison wrote: > "Richard Rasker" >> >> I regularly repair tube amplifiers, and there's one thing I occasionally >> encounter that doesn't appear to make much sense: a high resistor (> >> 100K) in series with one of the first 12AX7 preamp tube grids. >> >> OK, I fully understand that this resistor, together with the Miller >> capacitance between grid and anode, can act as a high cut-off filter, and >> that this can prevent HF oscillations and the likes, but it also >> introduces noise -- a LOT of noise. > > ** Not true, when it is in a later stage of the circuit than the input > stage. OK, but in this case, the resistor is found in series with the grid of the second triode -- and believe me, it's responsible for a lot of noise. > Cos the noise generated by the first tube stage swamps that resistor's > noise. > > BTW: > > 470kohms has a self noise of about 10uV in the audio band - much less in > the narrower band guitar amps use. > > The audio band INPUT noise level of a typical 12AX7 triode section is 3 > V - multiplied by the gain of 50 or more at the plate. >> I have here exhibit A, a Peavey 5150 EVH. Shorting the aforementioned >> 470K grid resistor in the second preamp triode cut the noise level by as >> much as 15dB(!). > > ** Not true, in normal use. I measured this with the "pre" gain turned 1/4 up and the "post" gain halfway up. In a silent environment, it makes the difference between audible noise and almost total silence. OK, at normal playing volume, it's hardly noticeable, but during quiet passages, it's rather annoying -- especially in a studio environment. >> So my question is if anyone can explain why any designer would use such a >> ludicrously high grid resistor > > > ** Simple - to prevent the previous tube stage from being asymmetrically > loaded by grid current when following stage is overdriven. That seems a bit far-fetched in this case, considering the placing in the circuit -- it's R82, behind the very first triode, with some filtering and the volume control in between. Anyway, I'll do some more measuring, experimenting and listening; perhaps the best thing to do is to decrease the resistor value to 100K or thereabouts. Thanks everyone for their input; I've learned a few new things about tube amp design :-) Richard Rasker -- http://www.linetec.nl
From: John Walliker on 12 Nov 2009 15:05 On 12 Nov, 19:12, Jan Panteltje <pNaonStpealm...(a)yahoo.com> wrote: > But really, with modern processing techniques you can make a lot more, and perhaps > better, effects at the low signal level and then play it via a good solid state amp and high > quality speakers... > Probably with more volume too... > It would be interesting to somehow characterise the transfer curves of some of those old tube things, > so it can be mimicked in FPGA or with a DSP for example. Many years ago (probably 1977 or 1978) Charlie Watkins of WEM gave a talk to my university audio society. One of the questions he answered was "why not use non-linear networks to simulate the valve sound" in guitar amps. The answer was that WEM had tried hard to do this, as there would have been considerable robustness and cost advantages. Unfortunately, they could not get the sound right, so they decided to carry on making valve amps. I think the problem was not just getting the right transfer function, but also the time dependencies like power supply sag and even valve microphonics that affect the overall sound under severe overload. John
From: Jan Panteltje on 12 Nov 2009 15:22 On a sunny day (Thu, 12 Nov 2009 12:05:16 -0800 (PST)) it happened John Walliker <jrwalliker(a)gmail.com> wrote in <c35f8d6c-f5dc-4916-9093-7c453e7cdfb5(a)r24g2000yqd.googlegroups.com>: >On 12 Nov, 19:12, Jan Panteltje <pNaonStpealm...(a)yahoo.com> wrote: > >> But really, with modern processing techniques you can make a lot more, and perhaps >> better, effects at the low signal level and then play it via a good solid state amp and high >> quality speakers... >> Probably with more volume too... >> It would be interesting to somehow characterise the transfer curves of some of those old tube things, >> so it can be mimicked in FPGA or with a DSP for example. > >Many years ago (probably 1977 or 1978) Charlie Watkins of WEM gave a >talk to my university audio society. One of the questions he answered >was "why not use non-linear networks to simulate the valve sound" in >guitar amps. > >The answer was that WEM had tried hard to do this, as there would have >been considerable robustness and cost advantages. Unfortunately, they >could not get the sound right, so they decided to carry on making >valve amps. I think the problem was not just getting the right >transfer function, but also the time dependencies like power supply >sag and even valve microphonics that affect the overall sound under >severe overload. > >John If you can describe it, then you can program it.
From: ChrisQ on 12 Nov 2009 17:06
Jim Thompson wrote: > > No transformers in the signal path. > Like all engineering, it's what gets the job done in the most effective way and a transformer is a perfectly respectable component to use for such an application, especially considering it's history of getting the job done well at low cost. It even adds weight to the box, so the roadies think it's been properly screwed together :-). Such stuff matters and you build the product, imperfections and all, to sell into that market. I got into micros in the late 70's, so you can see where my head was going. There's only so much you can do with a class ab amplifier and it was time to move on. If you could accurately model a Marshall 100 or Watkins AC30 or similar and then dsp the resulting signal chain, you can bet that some muso's would still say they could still tell the difference and to be honest, they probably could :-)... Regards, Chris |