C3g Aikido Phono

[simon]

I like silver mica for EQ

https://www.ebay.com/itm/254730247967?v ... R4a85vq_YQ

That's interesting, I read some fairly polarized views on silver mica in RIAA yesterday. I have some smaller versions of the Russian type in your link.

[simon]

It is a nominal 5k6, if memory serves, it was done in Spice, but the values will scale up, or down, according to requirements.

We did, at the time, think low Z had other advantages, so it was about striking a balance, but that's where my memory goes hazy.

It is a long time ago but my recollection also is that it's 5k6 impedance - half way between 600R and 10k ISTR.

[simon]

I've been around in circles a bit thinking about this yesterday and today. At one point I thought I'd rebuild the C3g and RIAA stage and see if I could improve things, but when I had a closer look at the existing I thought I probably wouldn't end up changing too much so decided against it.

Then I thought again that what's really needed is to get all the circuit together, at least per channel. Which led me back to the pcbs. But then the pcbs have to be cobbled together anyway, and they're not flexible to layout change.

So then I thought perhaps I need to rebuild the whole thing point to point. Which is a lot of work for a bit of hum.

Hmmmm.

[Nick]

It is a long time ago but my recollection also is that it's 5k6 impedance - half way between 600R and 10k ISTR.

I am not sure about that, I can believe that the network is correct (as it gives the correct frequency response) if driven from a 5k6 source, however that doesn't mean that the 5k6 source sees the network as a 5k6 load. Looking into the RIAA network, it looks to me that above 10kHz, the impedance of the 470nf gets down to 680R and below, so the load that the valve is looking into starts to tend towards 680R with increasing frequency.

If you build a two stage amp, its unlikely you would make the second stage grid resistor 680R, and if you did the 680R would at AC be in parallel with the 5k6 anode load so the load line you would draw for the c3g would tend to 680R from 5k6. Now being a pentode it will be ok driving into a low load but that's not to say the distortion profile of the result will not be the same as a 5k6 load.

[IslandPink]

I remember going through these deliberations when trying different low-pass filter values in my power amp, with the 4P1L pentode driver.
There was a stage when I was more or less relying on the anode load ( ~7.5k ) as the 'R' in the R/C filter, but Nick pointed out that the pentode would increasingly see a short as it went past the crossover frequency. When I re-calced it to include something like 7k resistor in-line as well as the anode load, it sounded a lot better.

[simon]

Okay I can see that I think. So, and without wanting to redesign Andrew's work, might there be a benefit in tweaking the RIAA? If it is scalable as above is it as simple as just scaling up the resistors and caps? And no, I'm not expecting it is!

[Andrew]

Yes, that RIAA will scale, the trick is choosing standard values, which is what that one does.

I did say it was a 'nominal' 5k6 as the pentode load confirms, but as Nick says, it will be lower at high frequency as the caps will be a 'short'.

PS apologies for stating the obvious but caps get smaller while the resistors get bigger.

[Nick]

But as mark did you can reduce the minimum load by keeping the 5k6 anode load but including a series r in the feed to the riaa.

[Nick]

Also if you use a bigger coupling cap you can put it before the riaa and then your riaa caps don't have any voltage requirements. All choices. Choices I too once made and since backed away from.

[simon]

What's obvious to one person isn't to another, thanks for mentioning it . I'll have a mull.

[Andrew]

But as mark did you can reduce the minimum load by keeping the 5k6 anode load but including a series r in the feed to the riaa.

Err, I'm a bit rusty, so I would need to think that thru'!

I wonder, out aloud, if you might need to increase the anode load on the pentode, if you did that? The pentode load matches the impedance looking into the network. Wouldn't it now need to match the impedance looking into the series resistor plus the network?

Regardless of that, one thing that my aging mind does remember is that one aim of that circuit, right or wrong, was to remove the need for series resistance?

PS there was also a good reason why I choose pentode too, if only I could remember it......

[Andrew]

Also if you use a bigger coupling cap you can put it before the riaa and then your riaa caps don't have any voltage requirements. All choices. Choices I too once made and since backed away from.

Also, it was originally designed to have the cap after the network, so it can be smaller, as Nick says, lots of ways to skin the cat.

I will add without any desire for brownie points since I was too lazy to build this and the credit should be Simon's, but this phono, on the day, was the closest, by general consensus, to Nick's LCR, whatever version of the LCR it was - I don't recall that either!!!

[Nick]

Well for example keeping the anode load at 5k6 but calculating the riaa to driven by 61k can be done by adding a 56k resistor in series before the riaa network. That way the pentode sees a more constant load and so distortion with Freq is more constant and the riaa is driven by a more stable load as the GM of the valve changes as it ages.

Also wrt the distortion against Freq with the original circuit not only does the driven load tend to 670r above 10k so the driving amplitude increases at 6db per octave above 10k doubling the valves problem.

I agree we was of the mind that lower impedance was better. Not sure if that's true anymore and certainly not true taken to that extreme. But hindsight is wonderful after all.

[simon]

Well, I was only trying out your thoughts Andrew to be fair.

I don't recall that comparison I have to say, but it is very nice sounding .

[Andrew]

Well for example keeping the anode load at 5k6 but calculating the riaa to driven by 61k can be done by adding a 56k resistor in series before the riaa network. That way the pentode sees a more constant load and so distortion with Freq is more constant and the riaa is driven by a more stable load as the GM of the valve changes as it ages.

Also wrt the distortion against Freq with the original circuit not only does the driven load tend to 670r above 10k so the driving amplitude increases at 6db per octave above 10k doubling the valves problem.

I agree we was of the mind that lower impedance was better. Not sure if that's true anymore and certainly not true taken to that extreme. But hindsight is wonderful after all.

Two questions..

Are you scaling the network too? I think I need to get my Spice crutch out

Doesn't a pentode's output impedance remain stable as it ages, I think that was one point of the pentode front end, well, at least one of them that we thought at the time. In fact I seem to remember modelling this....