You're the one making the claims. You're a very good troll I will give you that. You have a talent for pseudoscientific explanations of non-existent phenomena. You could make a lot of money writing books on quack medicine.
After graduating, I almost went to work for a company in New Jersey that manufactured military analog computers.
Guess that was quite a while ago.
So the demo card is new? The part has been around it looks like for 13 years.I just got a full demo card of our AD1955 DAC our ESS 9018 "equivalent". In their infinite wisdom it was laid out to facilitate op-amp rolling. So I'm ready to pick open the sores on alternative DAC filtering.
Does the 1955 have a spec on preserving performance requiring a maximum "current" output voltage burden? (lazy here).
Well I guess it was a popular use of the things for calculating trajectories and figuring out how to lob leaden death at targets. I liked the mechanical ones, that one would see in surplus.
No output compliance spec that I've been able to spot in the DS - just an injunction against using passive I/V 😛
Yeah. I remember Barrie Gilbert's polemical discussion of how bad conventional op amps should be for I-V conversion. Nelson says some like passive even when it may entail non-monotonic behavior. The one I saw that raised my eyebrows had, I recall, a 10 ohm terminating resistor, followed by a tube stage.
I thought it was cool to see the differential equations unfolding in real time.
In terms of audibility I reckon its because their OPSs aren't pure class A over the range of DAC output currents (8mA or so swing for the 1955). So perhaps a current source bias on the OPS would help. I've tried that on my own DACs (not AD1955 any longer but I was using one of Barrie's chips), not completely certain it made an improvement but I left it in anyway.
I was a fan of passive for many years, gone back to active again now - the PSRR benefits of active won the day. Better monotonicity can't but help though.
I agree. I simulated an oscillator circuit (differential equations) on the Heathkit analog computer back in 1962. I thought it really elegant to be able to do so. They were still using them full time at Lockheed in 1963, but I only ran the digital mainframes at the time. Somewhere, along the line, I was told that a digital program virtually replaced the analog computer, even with an IBM 1401 computer, and I guess they faded away from there.
Barrie Gilbert is a really sharp guy. IF he criticizes an IC op amp in some way, I am prepared to listen up.
The elephant in the living room is How much does the output Z of this so-called current source change as a function of code? I don't know any manufacturer who bothers to mention it. There are some lower-priced DACs that really do have rather high-Z outputs, but they don't seem to be very popular.
I think EUVL's simple JFET common-gate designs with floating power supplies are pointing the way to a more suitable approach. Making the input still lower Z with an added amplifier can manage very good performance. You're starting with low Z to begin with and just making things lower, rather than starting with high Z and using a lot of feedback to make it low Z.
But op amps are very convenient. And probably quite adequate.
I thought I had seen somewhere that the PCM1794 app note circuit with the NE5534s actually got some benefit out of the offset current due to that. The AD1955 datasheet shows pull-up resistors to supply that.
I worry more (coming from a SQ pov) about the magnitude of the Zout because lower Zout means poorer PSRR in my simple logic. Some very popular DACs have surprisingly low Zout (PCM1704 springs to mind @1kohm) and yeah, no mention of the degree of variation over code.
Schitt just released another DAC using a multiplying part (AD5547) - the DS says Zout is code dependent but leaves it to the reader to work out what it'll be from the internal schematic 🙂 The 200pF output capacitance is also code dependent.
Which are the cheap DACs you've found with high Z outputs?
Last edited:
- Status
- Not open for further replies.