It's time I built a D3a -> 5687 phonostage

[IslandPink]

If you drive it with matching Z you can terminate with a high Z, I think that approach is viable at 1k5, but 600R certainly makes things harder. To do this you need chokes that are happy at anode DC. A matching termination and those aforementioned huge caps avoid such issues and the sound quality of the solution speaks for itself.

What was that system Thorsten used in the first design, I forget what he called it ?
He had a big coupling cap , plus a small one in parallel. There was a resistor in series with one of the caps. I wonder if the bigger cap was an electrolytic ( BG ? ) . Did anyone understand via Spice, if that had some logic, when you took into account series inductance of the bigger cap ?

[Nick]

Don't know what he called it but the effect was hf went through the small cap and lf through the large one.

[Thermionic Idler]

I don't think I'm going to lose too much sleep over the cap coupling - the Spice model is suggesting that whilst the LF rolloff will occur earlier, it's still tolerable. I still have 4 big 47uF Obbligato film caps kicking around from an abandoned project years ago, I might press those into service.

The one concern I do have is matching the driving impedance to the lower impedance RIAA. 6C45P in place of the 5687 would certainly do the job as far as I can see, just the fun and games of making sure it doesn't oscillate (ferrite beads on everything?)

EDIT actually looks like the 5842 / 417A might be a perfect match, looks like about 2.5 times the transconductance of the 5687, and 1500/600 = 2.5.

[izzy wizzy]

It was called lossy parafeed not by Thorsten but by Voltsecond. I've used it to couple to TVC 1u film with 22u BG and 1k in parallel.

I thought it easier to terminate the LCR with the characteristic Z than work out the accurate drive Z unless driven direct off a pentode or similar. But then I don't understand cathode followers very well.

[Nick]

5687 CF is about 100R output impedance, and it does have feedback wrapped around it. I would check if its a problem before you try and find a solution. But YMMV.

[Andrew]

If you drive it with matching Z you can terminate with a high Z, I think that approach is viable at 1k5, but 600R certainly makes things harder. To do this you need chokes that are happy at anode DC. A matching termination and those aforementioned huge caps avoid such issues and the sound quality of the solution speaks for itself.

What was that system Thorsten used in the first design, I forget what he called it ?
He had a big coupling cap , plus a small one in parallel. There was a resistor in series with one of the caps. I wonder if the bigger cap was an electrolytic ( BG ? ) . Did anyone understand via Spice, if that had some logic, when you took into account series inductance of the bigger cap ?

Yes, he did, in his second iteration of the 600R LCR , there were some interesting design choices to make it work, like zero bias on the initial high gm pentode and the second D3A in grid leak, the later was a hangover from his more successful 'simple phono' that used an 88 in grid leak; the 88 seems more tolerant to this design choice, but I still think I wouldn't use grid leak again.

[Thermionic Idler]

5687 CF is about 100R output impedance, and it does have feedback wrapped around it. I would check if its a problem before you try and find a solution. But YMMV.

Ah yes, that's the feedback into g2 of the D3a isn't it, I should be able to work that one out.

[Nick]

Yes, one of the effects of the feedback is it will reduce the output impedance.

[Thermionic Idler]

Ha !
How about £25 each nowadays ?

I should have bought more

More than that, at least the two I've bought were. And this is a perennial worry because I didn't get in when they were cheap. I found some NOS Siemens for sale from a Netherlands supplier at a reasonable price that seemed perfect, but he refused to sell to me because of the red tape caused by Brexit (which brought back my Brex-rage for several moments).

I eventually got a couple from Italy but that was 9 days ago, still waiting for them to get here. The supply on eBay seems to be drying up. I'm starting to wonder if I should be doing this at all if one of the key valves is about to become unobtanium or hideously expensive. Can the circuit be done with current production? EF86? There probably isn't an alternative. All that along with wondering if I'm just biting off more than I can chew.

There's these PCB's from waltube on DIYaudio which were designed around the Sowter RIAA and use fairly common valves - he developed the things using much better test equipment than I have access to, but of course I haven't heard them, haven't seen any write-ups, so I just have no idea whether they're actually any good or not. But I've heard the ref phono a few times and if I can at least get most of the way towards what it does, then I'll be pretty happy.

Given what I'm about to cough up for Dom Harper's cartridge, I guess I can't moan too much about the price of D3a's...

[izzy wizzy]

D3a last for ages. In the end they are vfm.

[Thermionic Idler]

D3a last for ages. In the end they are vfm.

Rated to something like 10k hours IIRC?

[izzy wizzy]

Could try and find E810F but probably same issue.

[Nick]

Forget the number but there is a Russian high GM pentode that works as well. But I remember measuring some and finding two sets of curves as if there were two construction types.

[IslandPink]

6J52P ?

[Nick]

Yep, think that's it