5842 grounded grid

[JoshK]

I going on what I've read from people like Jocko Homo and others with a great multitude more experience in designing dacs then I. However, with a current output DAC, most are designed to ideally have a 0V (or in some cases a 3V) output.

In the case of the BB PCM63 for example, useage of >100R shunt resistor takes the DAC well out of its linearity range and distortion rises. Thus the passive I/V solution is suboptimal, however it is still widely used for its simplicity. The DAC's spec sheets also suggest one should present the dac with a low Z.

The best solutions, so I'm told, are transconductance amps, such as with GG, that amp the current and shield the DAC from the Z and keep the DAC's output near 0V. I'v read that this yields much better linearity from the take and yields a much better slew rate than an opamp solution. I am just interested in trying it out in the near future and have been thinking about circuits that will work with this idea.

[Sheldon]

Ok, more than meets the eye. It's not just low impedance then.

Sheldon

[Eli Duttman]

Josh,

I dislike raining on your parade, but a GG buffer for a current DAC could be a BAD idea. The reason is the same in trying to use a GG gain block with a LOMC phono cartridge. A considerable DC current is flowing in the cathode circuit. That current could try going around the cathode to ground resistor via the DAC chip. An ENORMOUS cap. is needed to block DC at these low impedances. YUCK!!!

You say a 100 Ohm I/V resistor takes the DAC chip out of its linear range. Use the largest value less than 100 Ohms that's consistent with good linearity as both the DAC I/V and grid leak resistor of a common cathode 6922 section. CCS load the triode and DC couple a source follower to its plate. Gain will be darn near 33X and the only current flowing in the I/P circuit is AC from the DAC chip.

[Brian Beck]

I dislike raining on your parade, but a GG buffer for a current DAC could be a BAD idea. The reason is the same in trying to use a GG gain block with a LOMC phono cartridge. A considerable DC current is flowing in the cathode circuit. That current could try going around the cathode to ground resistor via the DAC chip. An ENORMOUS cap. is needed to block DC at these low impedances. YUCK!!!

Or...you could make the first stage a GG with direct connection to the current-output DAC by using a cathode CCS returned to a low-voltage negative supply with a servo control to hold the input at zero volts. Use a couple of back-to-back diodes on the input to protect the DAC during turn-on and turn-off transients. Return negative feedback to the input to achieve a transimpedance amp. With a few 6DJ8 stages within a feedback loop you can approach 4 ohms of input Z which makes the DAC a lot happier than a 100 ohm resistor, and makes noise performance better. But you have to live with lots of feedback this way, essentially making a vacuum tube opamp.

[JoshK]

Originally posted by Brian Beck


Or...you could make the first stage a GG with direct connection to the current-output DAC by using a cathode CCS returned to a low-voltage negative supply with a servo control to hold the input at zero volts. Use a couple of back-to-back diodes on the input to protect the DAC during turn-on and turn-off transients.

This is what Broskie shows in his blog, and what I thought of doing. I hadn't explained that as I was more interested in the building blocks, but I guess it was necc to explain.

The part about the NFB isn't used in JB's approach. I'll have to think about it.

[Brian Beck]

Josh,

I haven't seen Broskie's design. Do you have a link to it? Another tube I/V design is an old one by David Berning, for which I have no link.

This is one place where feedback makes sense to me. Think of an opamp with non-inverting input grounded, the inverting input connected directly to the DAC, and a feedback resistor from the output to the inverting input. The DAC designers had this in mind while doing their design work, not a 100 ohm resistor. We'll just substitute tubes. With enough gain and feedback, there is no AC on the inverting input, but the AC signal current from the DAC goes through the feedback resistor. So the transimpedance gain of the amp is Vout/Iin = -Rfb.

[JoshK]

Here is the schematic and the link to the blog where he discusses this topic is here...

http://www.tubecad.com/2006/07/blog0072.htm

My naive thoughts, because I am still relatively new here, is to replace the 20K plate resistor with the top half of a mu stage. I will have to read and ponder some more about your ideas on turning it into a tube opamp.

[Brian Beck]

Thanks Josh,

Well, that’s simple enough. The problem is that the input resistance (the cathode) is still pretty high because he doesn’t use feedback. The cathode gives about 770 ohms, which is in parallel with the 60 ohm resistor, for a combined resistance of 55 ohms. The PCM63, for example, has a current output port resistance of 670 ohms, lower than you’d think. It expects to drive a load much lower than that (a virtual ground) to ensure accuracy of the D-to-A conversion. 55 ohms is still a large portion of 670 ohms, better than 100 ohms, but I would want lower still.

I now notice that Broskie calls out the TDA1541 for the DAC. Its output resistance is not specified in the data sheet, but the data sheet for the “A” version says this: “To ensure no performance losses, permitted output voltage compliance is ±25 mV maximum.” Since it puts out 4mA at peak, this means that the load resistance must be lower than 6.25 ohms for the DAC to work to spec. I expect this is typical for most current-output DACs.

[JoshK]

Thanks Brian for all the help and suggestions. I see your point.

[Eli Duttman]

Guys,

I've started a new thread that focuses on current O/P DAC chips, instead of GG topology.