Hi
Can anybody help with this please?
I plan to build a PCM1798 DAC using the attached circuit for I/V conversion if its input can be adapted for connection to the PCM1798's differential outputs. I have been using the circuit as shown with a TA1541A, without the FET output buffer shown in the diagram, driving the following amplifier via a stepped attenuator and like its sound very much, therefore if it can be adapted for differential input I would like to try it with the PCM1798.
Can this be done and if so what would the new circuit look like please? I hope this isn't a stupid question and thank you in advance to any help offered.
Tim
Can anybody help with this please?
I plan to build a PCM1798 DAC using the attached circuit for I/V conversion if its input can be adapted for connection to the PCM1798's differential outputs. I have been using the circuit as shown with a TA1541A, without the FET output buffer shown in the diagram, driving the following amplifier via a stepped attenuator and like its sound very much, therefore if it can be adapted for differential input I would like to try it with the PCM1798.
Can this be done and if so what would the new circuit look like please? I hope this isn't a stupid question and thank you in advance to any help offered.
Tim
Attachments
The differential current output is simply two current outputs in anti-phase. So you'd use two instances of your circuit per channel. Follow those with a differential-single ended filter stage if you want a single-ended output.
Thanks, that makes sense. I'll look for a suitable differential to single ended buffer circuit or chip. If you have any suggestions in that direction I'll follow them up.
If you are open to considering an alternative circuit, I suggest looking at the below linked discrete I/V of Sergio Santos, a contributor here at diyAudio. It consists of only 7 transistors, yet, is fully differential, exhibits very low distortion and a very low input impedance, which is good for a current DAC to see. All in a single stage without a follower. In addition, it appears to incorporate a first order output filter, formed by C2 and C6.
https://www.diyaudio.com/forums/dig...convertion-low-distortion-95.html#post5767587
The other big advantage of Sergio's circuit is that the two lower current sources can be set to also sink the D.C. Bias current from PCM1798, which I recall as 3.9mA. In the circuit you're considering, that bias current will cause a large imbalance in operating conditions and offset.
https://www.diyaudio.com/forums/dig...convertion-low-distortion-95.html#post5767587
The other big advantage of Sergio's circuit is that the two lower current sources can be set to also sink the D.C. Bias current from PCM1798, which I recall as 3.9mA. In the circuit you're considering, that bias current will cause a large imbalance in operating conditions and offset.
Last edited:
Ken
Thankyou for the suggestion. It looks interesting - simpler to build than four instances of my original circuit.
May I ask how the lower current source would be configured to sink the bias current, I assume a trimmer pot somewhere but that's the limit of my knowledge, unfortunately?
I also wonder about transistor substitutions. I've not use the BC337/327 before but can obtain them. I do already have some 2SA970 and its NPN equivalent which I like very much (subjectively) in other circuits I've built so could these be a straight swap for the BCs? Also, what might be minimum/maximum suggested rail voltages for this circuit?
Many thanks in advance for any advice and apologies for such basic questions as this isn't my particular comfort zone.
Thankyou for the suggestion. It looks interesting - simpler to build than four instances of my original circuit.
May I ask how the lower current source would be configured to sink the bias current, I assume a trimmer pot somewhere but that's the limit of my knowledge, unfortunately?
I also wonder about transistor substitutions. I've not use the BC337/327 before but can obtain them. I do already have some 2SA970 and its NPN equivalent which I like very much (subjectively) in other circuits I've built so could these be a straight swap for the BCs? Also, what might be minimum/maximum suggested rail voltages for this circuit?
Many thanks in advance for any advice and apologies for such basic questions as this isn't my particular comfort zone.
Hi, Tim,
Guessing only by the component values in Sergio's schematic, it looks to me like the supply voltage is roughly +/- 12V. The exact value required would depend on how much differential signal swing you require.
You can see in Sergio's schematic that is two posts earlier (#948) that he has the bias current set at 12mA. This can be increased to also accommodate the bias current sources by the DAC in a couple of different ways. For example, you could decrease the emitter resistors R14 and R15, or you could increase the bias string resistor R17.
As far as the transistors, I believe that they were chosen mostly for their low-noise. The 2SA970, 2SC2240 pair are also low-noise, so so they should be fine as substitutes. Anyhow, I don't see transistor noise as critical for an I/V because the signals are usually relatively large.
For better guidance on implementing his circuit, however, I suggest that you contact it's designer by sending a PM to Sergio. He is most helpful.
Hi Ken
Thank you, very helpful, I'll follow this up with Sergio.
I saw on another thread that your had switched to passive I/V, finding it sounds better with the pcm1794. This is something I am also considering with the pcm1798, though in my case it would need a following gain stage with single ended output. I realise the pcm1798 has lower current output and wonder how best to judge between I/V resistor value and gain wrt distortion and noise respectively. I understand the pcm1798 bias current would be cancelled using both non inverting and inverting opamp inputs for this which is one advantage of an opamp I suppose.
Thank you, very helpful, I'll follow this up with Sergio.
I saw on another thread that your had switched to passive I/V, finding it sounds better with the pcm1794. This is something I am also considering with the pcm1798, though in my case it would need a following gain stage with single ended output. I realise the pcm1798 has lower current output and wonder how best to judge between I/V resistor value and gain wrt distortion and noise respectively. I understand the pcm1798 bias current would be cancelled using both non inverting and inverting opamp inputs for this which is one advantage of an opamp I suppose.
The lower the I/V resistor, the lower will be the distortion of the DAC. However, I felt that I could not hear any subjective improvement in the sound below about 25 ohms for the PCM1794A, which works out to about 50 ohms for the PCM1798, the exact value is not critical. I did not find noise to be audible with any value of I/V resistor I tried. As a point of reference, I recall the Audio Research had a tube DAC featuring the PCM1794A which utilizes 200 ohm passive I/V.
After experimenting with active (op-amp) output stages, I discovered that the signal was great enough to drive my discrete solid-state preamp directly. My output signal is about 200mV peak differentially, via 25 ohm I/V. It would be the same via the PCM1798 utilizing 50 ohm resistors. The line output impedance of the DAC is determined by the I/V resistors. Which is 25 ohms single ended, and 50 ohms differentially for my PCM1794A DAC. The signal goes from the DAC chip, to the passive I/V, then out through RCA connectors via coupling caps. (BG 4.7uF bipolar 'N'). I also have a nano-farad sized poly-film & foil cap. in parallel with each I/V resistor. This produces a simple first order low-pass filtering of the signal, as I don't find that the sound subjectively benefits from additional filtering.
Should your preamp not offer enough gain to do this, just determine how much additional gain is required to drive your preamp, and build the DAC's active output stage to produce that gain. My suggestion is to not be concerned over optimizing the measured SNR. It's not worth the bother.
After experimenting with active (op-amp) output stages, I discovered that the signal was great enough to drive my discrete solid-state preamp directly. My output signal is about 200mV peak differentially, via 25 ohm I/V. It would be the same via the PCM1798 utilizing 50 ohm resistors. The line output impedance of the DAC is determined by the I/V resistors. Which is 25 ohms single ended, and 50 ohms differentially for my PCM1794A DAC. The signal goes from the DAC chip, to the passive I/V, then out through RCA connectors via coupling caps. (BG 4.7uF bipolar 'N'). I also have a nano-farad sized poly-film & foil cap. in parallel with each I/V resistor. This produces a simple first order low-pass filtering of the signal, as I don't find that the sound subjectively benefits from additional filtering.
Should your preamp not offer enough gain to do this, just determine how much additional gain is required to drive your preamp, and build the DAC's active output stage to produce that gain. My suggestion is to not be concerned over optimizing the measured SNR. It's not worth the bother.
Last edited:
Ken
Thanks again for the very useful information. I will try passive I/V with an active gain stage using various opamps including a DIY discrete version, alongside Sergio's I/V circuit to see which sounds best.
Tim
Thanks again for the very useful information. I will try passive I/V with an active gain stage using various opamps including a DIY discrete version, alongside Sergio's I/V circuit to see which sounds best.
Tim
Looking at the attached and excluding the output buffer. How is the scaling of all the resistors values, R1 to R6, achieved in order to accommodate different rail voltage please?
I'd like to try this circuit with +/-15 volt supplies and with 2SA970/2SC2240 transistors in place of those shown.
Any help would be much appreciated!
I'd like to try this circuit with +/-15 volt supplies and with 2SA970/2SC2240 transistors in place of those shown.
Any help would be much appreciated!
Attachments
Last edited:
As a very approximate rule of thumb, scale the resistor values by the ratio of the new power supply voltage to the old one. Which in this case is 15/9 or an increase of 1.66. Which gives R2 = 9k1, R3-R6 = 2k4. VR1 I'd increase to 22k.
On second thoughts on VR1, keep the 10k but put 6k8 in series with it. Then the adjustment of offset will be a little bit easier.
On second thoughts on VR1, keep the 10k but put 6k8 in series with it. Then the adjustment of offset will be a little bit easier.
Last edited:
Thank you, I thought that might be the case. Presumably swapping the transistors won't introduce another significant variable in this instance.
Have you actually simulated or measured this circuit with a TDA1541, or +/-2mA current signal ?
The input stage is biased at less than 2mA.
Even with push-pull in mind, I find that a bit low with the said signal current.
As an example, the circuit by Pedja for TDA1541 is biased at ~3.7mA.
https://www.diyaudio.com/archive/bl...-out-production-pedja-rogic-ddnf-iv-stage.pdf
Patrick
The input stage is biased at less than 2mA.
Even with push-pull in mind, I find that a bit low with the said signal current.
As an example, the circuit by Pedja for TDA1541 is biased at ~3.7mA.
https://www.diyaudio.com/archive/bl...-out-production-pedja-rogic-ddnf-iv-stage.pdf
Patrick
EUVL, that's interesting. Yes, I have this circuit working in two DACs, one with a +/-2mA signal using the 2SA970/2SC2240 combination. I have not run into any problems with it and it sounds superb but I don't recall the rail voltage or resistor values. The other is PCM56. From memory both DACs have different rail voltages from those shown here. Maybe I need to open them up! The circuit was simulated by a friend/guru who can unfortunately no longer help with this but this is what he sent with the circuit as shown in my earlier post.
"The current in the I/V stage is very much dependant on the Vbe of Q1 and Q2 and you may need to experiment with different values for R2 to get the Iq of this stage to around 6mA"
If I measure R2 and adjust for 6mA does anything else need altering to suit?
EDIT....I've just measured the resistors in a BC550/560 version I've been using on +/-15V rails and they measure 3k3 for R1/2 and 680R for R3/4/5/6. I had been wondering why they were lower, and with 15V rails, but perhaps it makes sense now, given what you have said.
"The current in the I/V stage is very much dependant on the Vbe of Q1 and Q2 and you may need to experiment with different values for R2 to get the Iq of this stage to around 6mA"
If I measure R2 and adjust for 6mA does anything else need altering to suit?
EDIT....I've just measured the resistors in a BC550/560 version I've been using on +/-15V rails and they measure 3k3 for R1/2 and 680R for R3/4/5/6. I had been wondering why they were lower, and with 15V rails, but perhaps it makes sense now, given what you have said.
Last edited:
PCM56 is +/-1mA.
PCM1798 is +/-2mA.
So I would use at least 2x the bias.
You can also have a look here :
A Simple Discrete Current-Mirror IV Converter, à la AD844
DAC AD1862: Almost THT, I2S input, NOS, R-2R
DAC AD1862: Almost THT, I2S input, NOS, R-2R
Cheers,
Patrick
PCM1798 is +/-2mA.
So I would use at least 2x the bias.
You can also have a look here :
A Simple Discrete Current-Mirror IV Converter, à la AD844
DAC AD1862: Almost THT, I2S input, NOS, R-2R
DAC AD1862: Almost THT, I2S input, NOS, R-2R
Cheers,
Patrick
Thanks again for you help!
I have measured the resistor values of this circuit in my PCM56 DAC and they read 1k2 for R1 and R2, and 680R for R3 to R6. With +/-12.5V rails how do I calculate the bias current as it stands?
I have measured the resistor values of this circuit in my PCM56 DAC and they read 1k2 for R1 and R2, and 680R for R3 to R6. With +/-12.5V rails how do I calculate the bias current as it stands?