Galvanic isolation from the sound card? Are we willing to live with the expense and/or nonlinearities of transformers? I wonder how much distortion would be introduced...
It would probably be possible to allow the unit to be built with or without them.
As far as I'm concerned, the right place for isolation would be the digital interface between the DAC and the computer. There are USB isolators around, but only for full speed, so it would limit the sample rates. There are isolated USB DACs around, but I've not yet seen an isolated unit with ADCs as well.
BTW, you're looking at $70 each for something like this: http://www.jensen-transformers.com/datashts/10kbdpc.pdf
Last edited:
I own one of the original interfaces. Had some problems with noise as well, but found careful (sometimes unexpected) grounding methods drastically improved performance.
Personally, having a transformer anywhere in the chain would defeat the purpose of high accuracy measurements, and I use iron in copious amounts in the signal path. I have seven between the D/A and the speaker. But no thank you in the measurement circuits.
Personally, having a transformer anywhere in the chain would defeat the purpose of high accuracy measurements, and I use iron in copious amounts in the signal path. I have seven between the D/A and the speaker. But no thank you in the measurement circuits.
Audio Precision used transformers in their generator outputs in the System 1 and some later models as well.
I'm using transformers on the send and return from the interface, this allows me to make measurements on input signals in the mV range without a lot of conducted noise from the PC power supply. I'm using WE REP-111C which provide good flat bandwidth out to at least 30kHz, and a slight bit of roll-off below 30Hz where the interface is also starting to roll off, in critical applications a little EQ compensates nicely for it. Yeah, they're probably rare and expensive at this point, but I suspect Lundahl makes something more than acceptable.
You could also use the coaxial spdif I/O on a 2496 or 24192 with spdif transformers and a Wolfson WM8804 or two (going discon soon) with an ADC and DAC of your choice.
I'm using transformers on the send and return from the interface, this allows me to make measurements on input signals in the mV range without a lot of conducted noise from the PC power supply. I'm using WE REP-111C which provide good flat bandwidth out to at least 30kHz, and a slight bit of roll-off below 30Hz where the interface is also starting to roll off, in critical applications a little EQ compensates nicely for it. Yeah, they're probably rare and expensive at this point, but I suspect Lundahl makes something more than acceptable.
You could also use the coaxial spdif I/O on a 2496 or 24192 with spdif transformers and a Wolfson WM8804 or two (going discon soon) with an ADC and DAC of your choice.
Try using an Tektronix AA501A analyzer front end circuit as a possible guide to protection ( basic clamp) with low distortion.
Brilliant! Literally... they used an incandescent lamp. 120V nominal with a cold resistance of 1k. Perfect!
That's a popular request and it would be nice, but I see no way to implement that without significantly compromising the noise performance. I would mean you need an input impedance of 1 megohm...
But where it's useful (e.g., plate circuits), the signal levels are pretty high. Where you have a 1x probe, the 1M is shunted by the source.
Right now, you're at 100k, so there's only ~7dB noise difference, worst case.
Right now, you're at 100k, so there's only ~7dB noise difference, worst case.
Based on the feedback I've gotten, adding 7dB of noise is unacceptable.
How about we just insert 900k in series with the BNC input (bypassed with a switch for normal unbalanced measurements)? Then a 1x probe becomes 10x, and a 10x becomes 100x. If you're measuring a plate circuit that should be fine, right? A 10x probe with the input set to 2V full scale would give you 200V full scale...
Pete
But that's worst case- no shunting by the input source and low signal level. If you implement the series resistor, what are the HF losses?
Trimmer, of course. 😀 What this doesn't do is buffer the circuit under test from the 150pF or more of test cable capacitance.
I found this project to interface with good specifications.
Suddenly one can avail any idea.
Audio test interface
Suddenly one can avail any idea.
Audio test interface
- Status
- Not open for further replies.