common-grid I to V converter


I'm looking for some schematics and opinions on the sollutions to use tubes directly after the TDA1541A DAC instead of the OP's.
I have seen some schematics and they are similar to Torsten Loech's design with just a small resistor as the I/V converter.
John Broskie proposed a common-grid converter but I din't really understand the negative CCS and would like to have first hand info from someone who have tried this.

It seems the DIY member ultranalog have tried this some years ago but his homepage is all about something else these days.

Anyone else?
I didn't try it but the CCS seems there to sink the polarization current of the tube without developing a DC voltage on the output node of the DAC, that should be a few mV from Gnd.
If you note the voltages on the anode resistor you have 100V on a 20k resistor, corrisponding to 5mA.

IMHO Broskie overlooked that there is also DAC current (2mA sink at idle so probably a 3mA CCS seems to be more accurate)

As for other questions, as the current efficiency (i.e. the % of the signal current that actually flows into the tube) I give the word to tube experts.


My understanding is that the input impedance of a common grid amplifier is the inverse of the transconductance (1/gm). This is, of course, dependent on the bias conditions, but for the 6n1p, assuming 5uS gm, that gives you ~200r input z. The DAC in Broskie's schematic sees both the input z of the tube cathode in parallel with the bias resistor. 6dj8 is a better choice for a common grid I/V converter, as it has a significantly higher gm. You can get 80r input z pretty easily (12.5uS gm). What I think would be really neat is an E810F triode connected and biased hot. This, as well as some other tubes (6c45p, WE437, 3A/167M, D3A, and a whole slew of other triodes and triode connected pentodes) have a transconductance nearly as high as some FETs, and can get the input z under 25r. You can also, as oshifis said, play games with input transformers. Still not close to what BJTs can get, but not too shabby, and I bet it could be made to sound GREAT.
Really thanks guys for taking time to answer :)

I guess I have to try it out myself but... I only have one CD player with " a real DAC TDA1541A", my referenceplayer I allready have done some tweaking at, and I'm afraid to fiddle to much with experiments and maybe get it broken.

I have some Philips NOS SQ E88CC that would maybe make it if I try though :) If you look at loading impedances the common grid should be a better choice then to use just a resistor and a SRPP totem arangement? Still, there is some reports on good sound from that as well...
I am also very tempted to another aproach. To use the first half of the OP as it is and tap the output from it and have a tubestage there. Then I have the first OP's filtering allready there and I can keep the OP's. Then I can also do A -> B tests with the new tube circuit against the all OP stage. My Marantz CD65II have option for two more phono RCA's at the back. There is also plenty of room at the right side for PSU and tubestages :cool:

The first OP stage also have a deemphasis FET switch. Is there any recordings out there that will use it? As I have understand they are very rare and you could probably disregard the hole circuit, no?

I have a struggle to decide what to test. You see my time on earth and to make all this DIY project is limited :bawling:

So far I have got very good tips and answers. I appreciate that. This is probably the largest and best forum for audio electronics on this earth. Thank you so much so far :angel:

Keep on posting please
No transformer please

Hello again
I forgot to mention. I am not interrested in a transformer approach. I will never even try it. Much to expensive and bulky.

I am maybe dumb but I only use transformers where I really need them, as a voltage transformer for PSU's.

I can also accept them when I build a Tube powerstage. Thats it. No transformers in any other place. :grumpy: (stupid old man...)


diyAudio Moderator Emeritus
2002-10-24 10:19 pm
The looking-in impedance at the cathode is approximately (rp + RL)/mu. That isn't terribly low for most practical situations. For example, if you use something like an ECC88 with a 25k plate load, the cathode impedance will be approximately 1k.

Worse yet, if you want a standard 2VFS output, you are stuck with a 2k load resistor, which is almost a guarantee of high second harmonic distortion.

Voice of sad experience here! I butchered a Magnavox 582 (back when you could buy them new) to no good effect.


diyAudio Moderator Emeritus
2002-10-24 10:19 pm
Most CD players are set up so that the max signal output (corresponding to all 16 bits being "on") is 2V. If you have a 25k plate load, for example, the max signal will be 50V for a 2mA D/A range. That's a lot of swing, and distortion will be higher than if you asked the same tube to swing 2V.

What might work is a very high current tube with a CCS to take up the excess (e.g., for a TDA1541, if the tube is running 20mA, the CCS would be 18mA), but now we're getting pretty complex.
SY said:
The looking-in impedance at the cathode is approximately (rp + RL)/mu. That isn't terribly low for most practical situations. For example, if you use something like an ECC88 with a 25k plate load, the cathode impedance will be approximately 1k.
This is exactly what I measured! And this is why I used an input transformer (that gives the SRPP stage a push-pull drive as an additional benefit). This way I get around 5 ohms impedance seen by the DAC and around 2V full scale output with very low distortion. The advantage of this approach is that I can drive the primary of the tranny also push-pull from two DACs, one is driven by inverted binary data.