For a number of years I've been happily using three home grown AD1865 dac units - all relatively simple circuits with passive i/v stages.
Whilst playing around with different i/v resistors, I noticed that PRIOR to soldering the resistors into the circuit, there's about 4k of resistance between the hot output and ground of the DAC.
This is the same for all of my units. On one it reads 4.1k on the right channel and 4.07k on the left channel. So obviously even if the i/v resistors are matched, the resistance seen by the DAC is not matched due to this anomaly.
Has anyone seen, or got an explanation for something like this?
Whilst playing around with different i/v resistors, I noticed that PRIOR to soldering the resistors into the circuit, there's about 4k of resistance between the hot output and ground of the DAC.
This is the same for all of my units. On one it reads 4.1k on the right channel and 4.07k on the left channel. So obviously even if the i/v resistors are matched, the resistance seen by the DAC is not matched due to this anomaly.
Has anyone seen, or got an explanation for something like this?
Manufacturing tolerances? A slight difference will simply change the channel balance by a tiny amount. Anyway, the 'cold' resistance at a chip pin may not tell you anything useful about the normal operation of the circuit.
I just checked on a balanced AD1865 I have - the left channel is 3.8k (for both +ve to GND and -ve to GND) and the right channel is 4.05k. So the difference can be somewhat frustrating.
For example, I soldered two perfectly matched 330R iv resistors to an unbalanced version of the DAC yesterday, only to find that when I measured the resistance seen by the DAC at post iv conversion, one channel was 296R, the other 301R.
This imbalance, whilst tiny at this stage in the gain chain, would presumably be magnified quite significantly post the preamp and power amp stage?
Is it better to use a trimpot as the iv resistor in order to match the channels perfectly?
What value trimpot would be suitable if we wanted an iv resistor of roughly 300R?
Something like this might do the trick:
500ohm Spectrol 25 Turn Trimpot - Jaycar Electronics
For example, I soldered two perfectly matched 330R iv resistors to an unbalanced version of the DAC yesterday, only to find that when I measured the resistance seen by the DAC at post iv conversion, one channel was 296R, the other 301R.
This imbalance, whilst tiny at this stage in the gain chain, would presumably be magnified quite significantly post the preamp and power amp stage?
Is it better to use a trimpot as the iv resistor in order to match the channels perfectly?
What value trimpot would be suitable if we wanted an iv resistor of roughly 300R?
Something like this might do the trick:
500ohm Spectrol 25 Turn Trimpot - Jaycar Electronics
If you want to match the resistors, feed the DAC with 10kHz wav from a transport and then adjust resistors for the same Vout.
Your (DC!) measurements prove that you have tightly matched channels - it's all good. You do not need to worry about anything.
Anyway, DAC will never see the pure DC resistance values you measured, especially if you use LCR filtering after I/V resistors, and even more if you use active filtering, i.e. OP amp / transistor stage / buffering...
I'd enjoy the music if I were you and forget about the whole thing...
Nick
Your (DC!) measurements prove that you have tightly matched channels - it's all good. You do not need to worry about anything.
Anyway, DAC will never see the pure DC resistance values you measured, especially if you use LCR filtering after I/V resistors, and even more if you use active filtering, i.e. OP amp / transistor stage / buffering...
I'd enjoy the music if I were you and forget about the whole thing...
Nick
... if you decide to trim the resistance for the exact Vout, do not leave the trim pots on a PCB -> they will introduce undesired inductance and capacitance to DAC Iout.
Nick
Nick
Thanks Nick - probably simpler just to enjoy the music, as you say!
I was planning to solder a trimpot between the tip and ring (hot/cold) of the RCA jack and use that as the i/v resistor, which might work...but then of course one is stuck with the sound of the cermet trimpot (whether that is a good or bad thing I don't know)...
I was planning to solder a trimpot between the tip and ring (hot/cold) of the RCA jack and use that as the i/v resistor, which might work...but then of course one is stuck with the sound of the cermet trimpot (whether that is a good or bad thing I don't know)...
Last edited:
that will not work well because I to V conversion will have to take into account the PCB track resistance, capacitance and inductance as the result of having the I/V resistor located far away from DAC Iout pin -at the RCA connector.
Nick
Nick
Does it have that much of an effect?
Normally I simply make my own cables and solder the resistor between the GND and TIP inside the cable....so it would be at least 15 to 20 cm away from the DAC chip!
In other words, are you saying the iv resistors should be as close to the DAC chip as possible?
Normally I simply make my own cables and solder the resistor between the GND and TIP inside the cable....so it would be at least 15 to 20 cm away from the DAC chip!
In other words, are you saying the iv resistors should be as close to the DAC chip as possible?
If you check the DS, you'll note that the absolute values of resistor values and the current output have a wide tolerance (30%) but that the voltage output is controlled to 1%. This reflects semiconductor manufacturing - absolute values aren't well controlled, but ratios can be. So the precise value of the on-chip I/V resistor is adjusted in manufacture to match the precise magnitude of the DAC's full scale current output and all is well.
To get your external I/V to give such good channel matching as the on-chip resistor allows you'd need to measure and hand-select (as you seem to be doing anyway).
I'd say put I/V resistors as close to the DAC as possible - but I've got 5 -10cm distance in my current set-up without any audible problems. I am using balanced I/V though - that helps somewhat in rejecting noise pick-up.
Last edited:
Indeed - I've long since given up the idea of using Z-Foil or Econistor Wirewound, you can never get the matching right without blowing a hole in your wallet, and decent welwyn or RN55 give 99% of the sound quality with 100% cost effective 'matchability'.
You're well ahead of me in hearing differences between I/V resistors - I'm still using 'junk' grade 0805 thick films....😛
Excatly - there's barely any difference ... it was a very expensive lesson to learn.
In fact, I think (well definitely with the AD1865) the difference is more about the R of the i-v resistor. 75R < x < 150R sounds great, but needs an additional gain stage, whereas with, say, > 200R, the sound is a little less detailed, but can drive a power amp with 500mV sensitivity
The problem with DIY audio is that there really is no end to the rabbit hole, and there definitely is no holy grail. Pity the commercial audio world makes a living out of claiming that every new piece of gadgetry is the holy grail....
In fact, I think (well definitely with the AD1865) the difference is more about the R of the i-v resistor. 75R < x < 150R sounds great, but needs an additional gain stage, whereas with, say, > 200R, the sound is a little less detailed, but can drive a power amp with 500mV sensitivity
The problem with DIY audio is that there really is no end to the rabbit hole, and there definitely is no holy grail. Pity the commercial audio world makes a living out of claiming that every new piece of gadgetry is the holy grail....
I think that the internal DAC architecture has a lot to do with whether the absolute value of the I/V makes a difference. With my TDA1545A and TDA1387s I find it doesn't much, if at all. I put that down to having quite a highish output impedance (I've not measured it though). Whereas the DS for AD1865 says 1.7k (30% tolerance!). I currently have about 30R, with a fair amount of gain from my AD605s.
Yes I agree there's no one holy grail, rather many routes to audio nirvana and the rabbit hole is always deeper than I've previously thought. That's the beauty of DIY for me though, its really a long way removed from the commercial world isn't it?
Yes I agree there's no one holy grail, rather many routes to audio nirvana and the rabbit hole is always deeper than I've previously thought. That's the beauty of DIY for me though, its really a long way removed from the commercial world isn't it?
yes, the I/V resistors should be mounted on top of the DAC chip if possible (if DIL)
Nick
Well just for kicks I got a cheap set of 330R resistors, put the highest value as iv on the channel which showed 4.07k, and the lowest value as iv on the channel which showed 4.1k and got a decent match on the DC reading.
Then, feeding the dac with 10Hz, 40Hz and 351Hz tones I found that even if the iv resistance is slightly mismatched (eg I just tested 332R on the left and 330R on the right), AC voltage for a 315hz test tone is identical!
At lower frequencies (eg 10Hz, 40Hz) there was a minor minor AC difference - clearly inaudible, but nevertheless interesting.
Oddly enough, at 315hz, even though the AC readings on both channels were identical, the resistance differences were even greater (the difference was 5R when playing music at this frequency!).
It almost seems that the DAC is engineered to give better channel matching for frequencies within the human hearing range...if that is even possible!
No. You will have a 0.1455dB channel imbalance all the way through, provided all your other resistors are matched to much better than 1.7%.lordearl said:
Assuming perhaps 4 gain setting resistors (2 in preamp, 2 in main amp) each of 1% tolerance you could already be 8% out if you are very unlucky, giving a channel imbalance of 0.67dB. In practice the Central Limit Theorem probably rescues you a little.
- Status
- Not open for further replies.