[KimLaroux]

Hi peeps. Been here years and that's the first thread I start... I need your help here, cause I seem to have no idea what I'm doing.
 
I built this circuit based on the SOHA II input stage:
 

 
The 15 K resistors are 0.25 W so they can't dissipate more than 0.5 W together. I am using the 25 V rail of the output stage.
 
To help myself figuring an operation point, I drew a load line. The line I ended with goes from (0 V, 11 mA) to (160 V, 0 mA). Is this correct, or have I done the math wrong?
 
So anyways, based on this load line and the maximum power the place resistors can dissipate, I chose to set the tail CCS at 7 mA. This drops about 52 V across the plate resistors, leaving the plates at around 108 V.
 
I powered the amp and adjusted the trimmers until the plates sit at 108 V. The CCS seems to be working fine, the adjustments react as expected. Verifying the cathode voltage, they sit at 3.2 V, which seems a bit low but still close to the expected voltage.
 
But it's not working. I have no audio output. Only noise.
 
Debugging the circuit, I noticed that I get audio for a few seconds at startup before it goes silent. I also raised the current... until I got audio. I begin to have an audio output when the CCS is around 11 mA, the plates at 77 V and the cathodes at 0.7 V. I honestly have no idea what is going on here. Anyone sees something I don't? Anyone can explain to me why I only get an output at this point, and not before?
 
( and I can't leave the circuit working at this operating point as it drops almost a watt across the plate resistors, twice their rating. I also doubt the PSU could supply this much current for long. )
 
Any help would be appreciated. I'm a noob, and apparently I got no idea what I'm doing, so I won't mind being sent back to tube biasing 101.
Thanks.

 

[nikongod]

You have a cathode CCS where you should probably have a cathode voltage source. 
I'm not sure that 7.5Kohms is an appropriate plate load for even 2 sections of (what tubes are these anyways?) in parallel. 

 

[holland]

Put a resistor instead of the CCS.
 
For clarification, what you have is not the SOHA 2 front end.  It's more like a "SSMH" with a CCS on the cathode.  I say "SSMH" because it's just parallel triodes, which has been used for ages before the SSMH.
 
1) remove the CCS and use a resistor.  It should work fine.
2) Put the CCS on the plate, and it probably should work fine.  I think.
3) Cut the grid connection, cut the plate connection, and you have the SOHA 2 front end, which is a common cathode amplifier.  This should work.
 
You're effectively shorting the signal, which is why you get nothing.  The noise is probably the noise floor of the amp, and probably originating from B+.
 
Edit:  Re: your load line question.  It should be 5.5mA, not 11mA.  The current through the CCS will be split across both triodes.  The load lines are per triode.

 

[wakibaki]

 
This is 100mV pk-pk in, split 50% across the 2 input resistors to represent an input pot.
 
With these numbers I get a gain of 8. Anode load resistors show as dissipating ~9.5mW each. There's a lot of current flowing in R10 in my circuit, ~4.5mA, the current in the tubes is about 1mA each.
 
Maybe you've got a build error?
 
You'll get twice the gain if you use the current mirror load as in the original SOHA II. There's no good reason for paralleling the tubes as you have them, unless you have a particularly demanding (high capacitance) load. You need to look at the CCS, because you shouldn't have nearly as much current in what is R10 in my diagram
 
w

 

[holland]

I sim'd the circuit as OP posted.  Getting 40uV on a 0.5V input.  It seems more like a design error.
 
Since it looks like you're doing a hybrid.  Just use the SOHA input stage (original one), which is a simple grounded cathode.  CCS on the plate.  The SS output buffer stage should have a very high impedance, normally.
 
 

 

[holland]

With these numbers I get a gain of 8. Anode load resistors show as dissipating ~9.5mW each. There's a lot of current flowing in R10 in my circuit, ~4.5mA, the current in the tubes is about 1mA each.
 
There's no good reason for paralleling the tubes as you have them, unless you have a particularly demanding (high capacitance) load. You need to look at the CCS.

 
Your CCS model is different.  You're loosing current through one of the transistors.  The ring of 2 should be to ground, and not split.  Anyhow, his voltages are correct for what he's measuring on the DC side.  I have a SOHA II with a B+ of 245V with a 10mA CCS.  5mA goes through each triode, so that part is correct for sure.
 
+1 on no reason to parallel, especially since it looks to be a hybrid, where the 25V is sourced from.

 

[wakibaki]

Quote:

 
Your CCS model is different.  You're loosing current through one of the transistors.  The ring of 2 should be to ground, and not split.  Anyhow, his voltages are correct for what he's measuring on the DC side.  I have a SOHA II with a B+ of 245V with a 10mA CCS.  5mA goes through each triode, so that part is correct for sure.
 
+1 on no reason to parallel, especially since it looks to be a hybrid, where the 25V is sourced from.

 
You're right, I should have taken the emitter of Q1 to ground, it doesn't make that much difference, tho'. I can't see your values, but there's some other conflict between our sims, because when I make the correction, I get 600mV pk-pk. Perhaps you could post a pic with legible values? Your trace looks a bit piecewise-linear, why don't you copy my spice commands, they'll allow you to do FFT's.
 

 
All I can see in terms of timing information in your sim is 0.000000 mS right at the beginning. It's customary to allow some cycles for the circuit to reach stable values but I can see your trace starts at 0V, and with 5 cycles there before the 1mS marker appearing, you could be running well above the amplifier's upper frequency limit.
 
w 

 

[KimLaroux]

Thank you everybody for taking the time to study my circuit and give me your opinions. 
 
Quote:

You have a cathode CCS where you should probably have a cathode voltage source. 
I'm not sure that 7.5Kohms is an appropriate plate load for even 2 sections of (what tubes are these anyways?) in parallel. 

 
The SOHA doesn't have a voltage source there... unless I'm missing something. The cathodes DO sit at their expected voltage, after all.
I'm thinking a higher value plate resistor would improve PSRR, but then it would either drop too much voltage and/or limit the amount of current I can push trough the tube. My HV is only 160V after all. Any recommendations on what would be a better value?
 
Quote:

Put a resistor instead of the CCS.
 
For clarification, what you have is not the SOHA 2 front end.  It's more like a "SSMH" with a CCS on the cathode.  I say "SSMH" because it's just parallel triodes, which has been used for ages before the SSMH.
 
1) remove the CCS and use a resistor.  It should work fine.
2) Put the CCS on the plate, and it probably should work fine.  I think.
3) Cut the grid connection, cut the plate connection, and you have the SOHA 2 front end, which is a common cathode amplifier.  This should work.
 
You're effectively shorting the signal, which is why you get nothing.  The noise is probably the noise floor of the amp, and probably originating from B+.
 
Edit:  Re: your load line question.  It should be 5.5mA, not 11mA.  The current through the CCS will be split across both triodes.  The load lines are per triode.

 
I've thought about replacing the whole CCS with just a resistor - seems so much easier. I may end up doing just that.
On point 3, you mean to cut those on one triode while leaving the other triode connected?
Can you clarify where/how the signal is being shorted? That may be the problem, I just don't see it.
 
Quote:

I sim'd the circuit as OP posted.  Getting 40uV on a 0.5V input.  It seems more like a design error.
 
Since it looks like you're doing a hybrid.  Just use the SOHA input stage (original one), which is a simple grounded cathode.  CCS on the plate.  The SS output buffer stage should have a very high impedance, normally.
 
 

 
Yeah it's safe to say my design is worthless at this point. I'd just like to know why, and if it is fixable in any way.

 

[wakibaki]

Quote:

Yeah it's safe to say my design is worthless at this point. I'd just like to know why, and if it is fixable in any way.

 
It's a starting point. Why don't you get a copy of PSpice, if you haven't got one already, and frig around with the design in that. It can be very enlightening. Just google dmtriode.inc and download it from duncan's amp pages. IMO the design only needs a bit of tweaking of the CCS to produce a stable, if not spectacular gain.
 
I repeat my suggestion that you may have a build error.
 
w
 
Can you see any more errors in my sim schematic?

 

[holland]

Quote:

All I can see in terms of timing information in your sim is 0.000000 mS right at the beginning. It's customary to allow some cycles for the circuit to reach stable values but I can see your trace starts at 0V, and with 5 cycles there before the 1mS marker appearing, you could be running well above the amplifier's upper frequency limit.
 
w 

 
I'll pull it in later.  I just installed LTSpice to do this, and banged it together quickly.  I don't normally run my personal projects on my work systems.  The time won't matter, as it's an ideal voltage source, but I'll shift it.
 
Quote:

Thank you everybody for taking the time to study my circuit and give me your opinions. 
 
 
The SOHA doesn't have a voltage source there... unless I'm missing something. The cathodes DO sit at their expected voltage, after all.
I'm thinking a higher value plate resistor would improve PSRR, but then it would either drop too much voltage and/or limit the amount of current I can push trough the tube. My HV is only 160V after all. Any recommendations on what would be a better value?
 
 
I've thought about replacing the whole CCS with just a resistor - seems so much easier. I may end up doing just that.
On point 3, you mean to cut those on one triode while leaving the other triode connected?
Can you clarify where/how the signal is being shorted? That may be the problem, I just don't see it.
 
 
Yeah it's safe to say my design is worthless at this point. I'd just like to know why, and if it is fixable in any way.

 
The SOHA 2 doesn't run it that way because the tube topology is entirely different.  It's configured as a long tailed pair, running as a signal splitter, with gain.  It then recombines the signal with the current mirrors on the triode.  Note that the second grid is connected to ground, and not replicated.  The plates are also not tied together.  The current mirror combines the signals into a single ended signal.  That's basically my point #3.
 
You're running the tube as paralleled triodes.
 
Use a resistor, if you get sound then you know that's the problem.  Just note the load lines, the current you calculate through the CCS is split across the triodes.
 
Look at the SOHA (not SOHA II).  That's really want you want to do.  You want the CCS on the plate.

 

[KimLaroux]

I kind of just picked some designs on the web, studied them and then tried to match them to the hardware I had. Seems it's a bit more complicated than that...
 
I actually don't have any windows machine on hand, and the simulators available for Linux are useless for such circuits. I'll have to check if my school has them installed on the computers in the Electronics department. Thanks for the suggestion. 

 

[holland]

Quote:

I kind of just picked some designs on the web, studied them and then tried to match them to the hardware I had. Seems it's a bit more complicated than that...
 
I actually don't have any windows machine on hand, and the simulators available for Linux are useless for such circuits. I'll have to check if my school has them installed on the computers in the Electronics department. Thanks for the suggestion. 

 
LTSpice works with Wine.
 
FWIW, I removed the CCS, and used a 1K resistor and the signals amplify.  It's not ideal, but it works.
 
You could use a CCS on the tail, but you need to pull the signal from the cathode.  It becomes a unity gain buffer, there are designs like that out in the wild.  For an amplifier in grounded cathode, you really want the CCS on the plate.  For another configuration, take a look at the Torpedo amp.  CCS on the plate.  Transistorized bias.

 

[KimLaroux]

So why does it work in the SOHA? Is it because the tail CCS is hooked to the negative rail?
 
I'm still confused as to how it can work with the CCS connected to a -15 V rail. Since the input signal is grounded and not connected to that same negative rail... it's confusing. 

 

[holland]

Quote:

So why does it work in the SOHA? Is it because the tail CCS is hooked to the negative rail?
 
I'm still confused as to how it can work with the CCS connected to a -15 V rail. Since the input signal is grounded and not connected to that same negative rail... it's confusing. 

 
No, it's the circuit topology that doesn't allow the CCS on the cathode, it is not the CCS by itself.  It's where you put it.  I thought I explained it up above in post #10.  I am not an expert, these are just the things I learned.  I, perhaps, am not explaining it well.  Let me rethink on how to dispense the information in a more consumable manner.
 
Anyhow, here's the CCS on the plate. 
 
Edit:  BTW, this is what I have on my SSMH.  A CCS on the plate (a CRD) and a resistor.  You should recognize this part.
 

 

[holland]

Quote:

 

 
Can you see any more errors in my sim schematic?

 
I noticed with the 12AX7 it shows gain.  With the 12AU7, it does not.  I used the 12AU7 not the 12AX7.  It might have something to do with the bias and the CCS operating point.  I'm not sure.
 
Edit:  There seems to be something wrong with the 12AX7 config.  The AC is seeping through the CCS.  It's almost as if the CCS is not operating.