audio-talk

RhythMick wrote: ↑Tue Aug 24, 2021 8:43 pm Going out on a limb, possible someone will correct me. I believe R1 is the series total of the output resistance of the previous stage and the actual resistor. So if the calculators show R1 as 24K and the output resistance is 3K then you need a 21K series resistor.

Zout of previous stage calculated as anode load parallel with the valve plate resistance I think?

Be interesting to see how close I am...

Yep, I think that’s it. I need to get a bit more into the idea that I might actually have some idea what I’m doing.
As Douglas Adams said “Don’t Panic”

izzy wizzy wrote: ↑Tue Aug 24, 2021 9:06 pm
That will be very close. But there's other rolloffs in there. The varying load of the eq on the first stage, the load of the 2nd stage grid varying with the coupling cap, decoupling cap Z changes reflected into the anode are some others. I also compensate for the output stage coupling.

Depends on how accurate you want to be. It's surprising the effect of very small changes on aspects of timing and stuff you wouldn't think the EQ has. I go for as accurate as possible which sounds silly so try to get the theory within a 0.1dB window at least.

I’ll be grateful to get something that works.

That circuit will get you going really well.

I would only change a few things and allow space for a few others. The D3a grid stopper could be increased to the 470R level or so. I use 680R. The g2 cap may need to be increased to 22u.

At some point you might like to bias the 1st stage with LED or other diode. This is where you can change it to suit your taste.

Layout is key. IMHO a very very small 0V is an advantage. If you're more comfortable with a bus, then as short as possible. Use signal wiring from the sockets to the bus rather than taking the bus to the sockets.

And most important, keep the grid circuits around the D3a very small and if possible close to a ground.

Build close to a screen either on pcb material or freestyle close to copper foil. Mine is on pcb and in a wooden box so no need to be drilling metal yet to get it quiet.

Received the D3As from Nick this morning, so I've spent a whole day working out the circuit: God that was painful!
I've left off the cathode follower for the moment. At least we have a practical starting point now.
A couple of points to ponder:
With the 100n coupling cap where it is, the eq is exposed to the max HT on startup unless an IDH valve rect is used.
I may decide to move the coupling cap and 1M grid resistor to the left hand side of the eq to prevent this possibility from occurring.
Also the voltage ratings for the eq caps will be much relaxed if I do that. Not decided yet though. I'll have to look at what's available and the ultimate cost implications if I leave the cap and grid res where they are.
The coupler will have to be increased in value if I move it, so it can't throw off the eq, maybe 2.2u or higher.

400v caps not expensive. I'd keep it there and avoid raising the value.

It all looks good. Personally I'd halve the EQ Z to make the values easier. But the trade off you mention can be real and they may take some finding.

Keeping the coupling cap to 2nd stage where it is best. If I could get at our container, I'd just send the parts so depends on how much more it's delayed and how fast you get going.

Question - should the 22uF from D3a g2 go to ground rather than cathode ?

No, not if you want to stabilise G2 as per the datasheet curves, which are referenced to cathode as zero.

There are arguments for both ways. The ideal is to keep a constant g2 voltage, but what constant means to the valve is the voltage between its cathode and g2. Its knows nothing about gnd. However in theory the cap across the cathode resistor means that the cathode and gnd are at the point for AC at least. And I can see how taking the cap to the cathode instade of ground will allow B+ noise a route through top the cathode, however the cathode is decoupled to gnd so any invected noise should pass to ground via the cathode bipass cap.

When I tried it, I wasn't sure I could tell a difference, but it may depend on b+ noise (maybe).

Nick wrote: ↑Wed Aug 25, 2021 11:15 pm it may depend on b+ noise (maybe).

Talking of which Steve, the offer of my spare Maida HV regulator board still stands if it's of interest? It's the same circuit as the Neurochrome one, though a different PCB layout.

https://neurochrome.com/products/21st-c ... -regulator

smd components are already soldered up.

Nick wrote: ↑Wed Aug 25, 2021 11:15 pm There are arguments for both ways. The ideal is to keep a constant g2 voltage, but what constant means to the valve is the voltage between its cathode and g2. Its knows nothing about gnd. However in theory the cap across the cathode resistor means that the cathode and gnd are at the point for AC at least. And I can see how taking the cap to the cathode instade of ground will allow B+ noise a route through top the cathode, however the cathode is decoupled to gnd so any invected noise should pass to ground via the cathode bipass cap.

When I tried it, I wasn't sure I could tell a difference, but it may depend on b+ noise (maybe).

Understood thanks

Ray P wrote: ↑Thu Aug 26, 2021 7:03 am
Talking of which Steve, the offer of my spare Maida HV regulator board still stands if it's of interest? It's the same circuit as the Neurochrome one, though a different PCB layout.

https://neurochrome.com/products/21st-c ... -regulator

smd components are already soldered up.

I’d completely forgotten about that.

Cressy Snr wrote: ↑Thu Aug 26, 2021 1:11 pm I’d completely forgotten about that.

Just let me know if you want it Steve - if you tell me your voltage and current requirements I can use the calculator to configure it to suit - assuming 320V and a load of 120mA (as per your schematic above) you'll need a 300V secondary on your transformer, around 75-80VA. Not too much heat generated either.