• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Sanity check differential amplifier for DAC

Hi Chris, slow reaction here, but the value of that resistor seems to be a debated topic of itself. The inspiration for this DAC came from JE labs that uses an even higher 90 ohms, but it’s definitely on my list to vary and see how it alters the sound. Especially in balanced mode I should have twice the voltage so lowering it should be viable, thanks for the tip!
 
Decided to build a standalone test for the 10m45s, to see if it's working before I connect it to the valves, and it feels like I'm missing something simple here:

Screenshot 2025-01-03 at 15.42.32.png


I've rectified a 2*18V transformer with 4 diodes and 4700uF to get ~57V DC, however when I connect as above I don't get any voltage across R2, can anyone enlighten me and tell me if I'm doing something wrong? (For now there is no ground connection, could the fact that this 57V is "floating" be an issue? Also I've been wondering if I can just connect the + to the ground at a later stage?)

I've tried 3 different 10m90s, but all with the same result, they came from China although the seller got good reviews. For the last test I used a Vero board as I read they are sensitive to heat, so wanted to remove that from the equation.

Any thoughts are appreciated! thanks - Matthew
 
Ok, I ended up building the "classic" long tail pair in the interim. This is roughly what it looks like:
Screenshot 2025-01-29 at 19.54.42.png

Ra = 47k

The voltages roughly match the simulation, i.e. the B+ is roughly the same, similarity the plate and cathode voltages.

The problem, is the voltage at the grid. I would expect it to be the same as below the cathode resistor Rk, which matches the simulation. However when I measure the voltage at the grid, I only have 68V there, instead of the ~93V I would expect.

What could be causing this 20-25V drop? I would suspect very little current to flow from the grid via the 1 Meg capacitor to the bottom of Rk. Do I have a faulty valve or is there something wrong with the schematic/build?

I'm a bit reluctant to put in a different set of valves before understanding whether this is expected or something in the schematic is just wrong. Any thoughts appreciated!
Thanks!
Matthew
 
It could be the input resistance of your voltage meter, perhaps.

The input resistance of a digital multimeter or of an oscilloscope with passive probes is usually in the 1 Mohm to 10 Mohm range, which would cause a substantial voltage drop across the 1 Mohm grid resistors. If it's a digital multimeter, you could try measuring the voltage between grid and cathode rather than the voltages to ground.
 
I think an update for all those sharing their valuable advice here is warranted, I've got the circuit up and running now using two TDA1541s in balanced mode. Overall it seems a bit of an upgrade sound wise from running a single TDA1541, but I think I'm doing the schematic injustice because I've only breadboarded it on a vero board which arent know for their sonic added values.
Next step is to move the whole balanced circuit to a PCB and bring this a bit closer to the output valves to minimise the effect of long wires, I seem to pick up some hum because of this, although it's not disturbing once music is playing, especially not on speakers.

At the same time while I'm redoing the digital circuit I'm also thinking of adding a second valve rectifier to create a negative B+ (a B- I guess 🙂 ), I have some additional chokes and caps lying around, so I could create a rather symmetrical supply and use a fairly large resistor under the large tail pair as I'd need to drop some 350 volts to bring the grid at zero. With symmetrical I mean, same CLC filtering on B+ and B- (so two CLCs for the left and two for the right filter)

Would in your opinion such an approach make sense, or would I be better of building a CSS under the LTP?

Cost is not a consideration, I've got the chokes lying around, parts for the CSS I would have to buy (given my previous experience with a "cheap" supplier, I'll go for the real deal in that case), but this shouldn't break the bank.

As always any input much appreciated!
 
Thanks @Merlinb even more topologies to consider 🙂
To all, having thought about this a bit more, in a way by having a large B- and a large resistor to drop the voltage I'm effectively creating a sort of CCS the long way round.
I'm trying to circumvent the coupling capacitors connecting the DAC to the grids, but with such a large voltage to drop I'm not sure how stable the bias point of the cathode would be (compared to the grid I would only need to be a few 1 or 2 volts negative), the other consideration being whether a penthode (or 10m45s) could provide technically and/or sonically preferred solution.
 
OK, this is the schema I'm thinking of:

Screenshot 2025-03-10 at 22.45.32.png


I have a couple of concerns, which hopefully can be tested before I blow up a couple of valves or TDA1541 chips 🙂

First at start up, if the B+ and B- "rise" before the heathers of the 6SN7s heat up, would the +350 to -350 i.e. 700V be too much for the 6SN7GTBs? or would this be no issue as no current is flowing? I can delay the heat up of the rectifiers, but would prefer to keep it simple with that delaying the B+ and B-.

Second the 50 Ohm I/V resistors now also act as grid leak resistors, would such a low value be a concern? I've read a out maximum values but not minimum values for grid leak resistors.

As for the drift of the bias point I was taking about, this is the 4.7V at the cathodes, this could swing, but given I use the same transformer and same rectifiers, this risk of this should be minimal, especially if I make sure it's far enough away from the grids which are at roughly zero volts.

Anything else that anyone spots that could be improved, more than open to suggestions!
 
First at start up, if the B+ and B- "rise" before the heathers of the 6SN7s heat up, would the +350 to -350 i.e. 700V be too much for the 6SN7GTBs?
Oof that's not good, you would have 350V between grid and cathode, you'll pop the grids for sure. Normally you would add protection diodes between grid and cathode, but that means some fault current might come from the DAC, so depending on the value of R3 maybe that could be a problem for the DAC? I dunno. Perhaps it would be better to put a neon lamp between the cathodes and ground.

Second the 50 Ohm I/V resistors now also act as grid leak resistors, would such a low value be a concern?
No problem.

As for the drift of the bias point I was taking about, this is the 4.7V at the cathodes, this could swing, but given I use the same transformer and same rectifiers, this risk of this should be minimal, especially if I make sure it's far enough away from the grids which are at roughly zero volts.
Cathode voltage is self adjusting, you only need to think about it when using a CCS that might drop out.
 
Last edited:
The 50R I/V resistors are a tradeoff between generated voltage (more R is more better for voltage generated) and an ideal 0R I/V termination (zero is design center, so less R is more better). DACs can be thought of as a lot of parallel current sources being rapidly injected into a common current sink, and summed to make that instantaneous sample's current sum. Non-zero R loading of the current summing junction causes an error in the generated voltage, a result similar to crossover distortion.

I would suggest that, although in fashion, a large-ish resistor is not a low distortion load for a DAC. But hey, if you're happy, I'm happy.

All good fortune,
Chris