PCM1794 strange current output

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The data sheet of PC1794A says that the current of each 4 ot the output is 6,2mA. I've just measured the current output and I've found 1 chipe that has all 8mA and another one has 2 output with 7,45ma and 2 output of 8,1mA
Strange I thought. But the better is yet to come....
When I've tried to join two output pins into a differential fashion the current didn't double ! Infact I've measured about 11mA

So I'm wondering and asking to you why the corrent haven't doubled when I've joined two pins ?

thank you in advance
Aiace
 
The outputs of the PCM1794 and 1792 are held at a constant -6.2mA. This is what they mean in the datasheet when they list the BPZ current (bipolar zero current) as -6.2mA.

The outputs then swing +- 3.9mA around that value to give you your output swing. That is the output will swing between -2.3mA to -10.1mA.

The outputs of the DAC are also expecting to see zero load impedance.

If you use the analogue filter circuit as proposed in the datasheet, the first two opamps perform current to voltage conversion at a rate as determined by the value of the feedback resistor.

That is, if you're using a 750 ohm resistor 750x-6.2mA will give you a constant DC offset after the first opamps of -4.65 volts for both the +ve and -ve output halves of the DAC.

The next opamp is configured as a difference amplifier, as the DC offset is common to both the +ve and -ve outputs it is cancelled out leaving you with negligible DC offset. The analogue signal on the other hand is equal but opposite for the +ve and -ve halves so they sum constructively leaving you with the analogue signal.

The resistor connected to pin 20 at a value of 10k to ground, is necessary for biasing the internal output circuitry. If this value is different, IE larger, the output current will drop. I would not recommend making it any smaller as the DAC could explode.
 
I meadured it at zero BPZ usign a digital multimeter stright to the PCM1794A pins and nothing else is connected to them. The manual of multimeter writes that it has a voltage drop of 5mV/1mA, so it should have 5 ohm of resistance. I suppuse this is low enaught.

Strange.....any idea?
 
I think you need to connect the pins on the outputs so that they are allowed to sink current properly. If you measure the idle current consumption of the DACs analogue 5v supply I think you will find that it differs greatly depending on if the outputs are connected up properly or not.
 
Administrator
Joined 2004
Paid Member
I just popped in and fixed the thread title.

5th element is correct, and I suspect that the resistance of your meter is also high enough to cause some measurement errors - also I would not expect very good accuracy from generic DMM at relatively low current levels - do you know the specified accuracy of your meter?
 
Do you have a 20mA scale? When was the last calibration done?

Anyway, the datasheet specify +/-2% error for DC center current - that is not related at all with the DAC accuracy in AC. Adding the 5-10% from your meter you can get impressive numbers. Try measuring those mA in both directions, see if you get equal numbers (one positive, one negative). That will tell you how accurate the mesurements are on your meter.
 
Last edited:
scale is 40mA; the tester is new and just calibrated.

Frankly I don't think this issue related to the accuracy, simply because there are differences from DAC to DAC. Data sheet writes 6,2mA on the coloum of typical values whitout any tollerance, so I think this value can change a bit from DAC to DAC. But thi is not a real issue because, as far I've measured, any differtial output pins has the same output current.
I've also measured the voltage on the 10k Iref resistor and olso this is slight different.

But what is very strange to me is that adding 8+8mA, I get 11mA !!!! :scratch:
 
Last edited:
The datasheet allows for some 2% error on median value. Adding the two values directly is not possible, the courent sources are not ideal ones, you end up with one generating a sighty higher voltage drop on your meter and blocking the output of the other.
The way is supposed to be used that DAC is with zero impedance on outputs - that is achievable only with an OpAmp as I/V convertor, not with a resistor like in your case and everybody that goes that route. Even the small resistance (I think it is some 400mV/40mA=10 ohm on that specific scale) of your meter is too much.
I guess your measurements prove just that.
 
The datasheet allows for some 2% error on median value. Adding the two values directly is not possible, the courent sources are not ideal ones, you end up with one generating a sighty higher voltage drop on your meter and blocking the output of the other.
The way is supposed to be used that DAC is with zero impedance on outputs - that is achievable only with an OpAmp as I/V convertor, not with a resistor like in your case and everybody that goes that route. Even the small resistance (I think it is some 400mV/40mA=10 ohm on that specific scale) of your meter is too much.
I guess your measurements prove just that.

I agree with you but 5ohm(the internal resitor of the multimeter) multipied for 12,4mA means 62mV!!! Try and measure the offeset of yuor op-amp you'll be surprised to measure also worst values.
No 62mV can't justify the difference of current and absolutelly can't justify the 8mA of output that means 33% more of the typical value
 
You don't have 5 ohm. You have 400mV(base voltage scale)/40mA(chosen amps scale)=10 ohm.
And the offset of the opamps has nothing to do with the capabilities of the internal current source in the DAC. That source is not ideal, will have an internal resistance that put in series with the meter will generate the error.
Also, look in the datasheet for the "bipolar zero error" - that is +/- 2%.
Now, I don't have a PCM1794 to test, that is my educated guess. It might be that you have a defective one.
I thought 7.8ma was the current output??
No, that vaule that you are talking about is 7.8mApp - that's the maximum current swing.
Different from the Center Current at BPZ - that is -6.2mA +/-2%.
 
Last edited:
You don't have 5 ohm. You have 400mV(base voltage scale)/40mA(chosen amps scale)=10 ohm.
And the offset of the opamps has nothing to do with the capabilities of the internal current source in the DAC. That source is not ideal, will have an internal resistance that put in series with the meter will generate the error.
Also, look in the datasheet for the "bipolar zero error" - that is +/- 2%.
Now, I don't have a PCM1794 to test, that is my educated guess. It might be that you have a defective one.

No, that vaule that you are talking about is 7.8mApp - that's the maximum current swing.
Different from the Center Current at BPZ - that is -6.2mA +/-2%.

neither, I've got a defective "two":p
infact both of them give me strange values.
Now I'm wondering if the tens of iv converters you can find on the web (resistros, transformer, resistor + tube, bjt, and so on) that don't use an op-amp are wrong. As far as I knew you must avoid tha the voldage is low enoght to be small respct the 0,6V of proteciton diode. I think that 62mV (or 124mV if you like) can't make a 33% of error in anyway.
 
I doubt your PCM1794s are defective.

I would hook up a standard opamp I/V converter to the outputs and see what happens. The PCM1794 is designed to work best in this application, it is not designed to work with passive I/V conversion. Yes I've seen people doing this but it's always a fudged effort to keep the R low enough such that the DAC linearity is not degraded too much but still provides enough output.

Arda Technologies - AT1401 Flagship Audio D/A Converter

Is an interesting chip and is one I am waiting to be released, it has been designed to work well with both passive and active I/V.
 
Administrator
Joined 2004
Paid Member
I doubt your PCM1794s are defective.

<snip>

I'd second that, the internal current generators are operating on a 3.3V rail and this chip uses bipolar technology IIRC, I suspect voltage compliance is quite limited which doesn't normally matter since it is designed to work into a virtual earth. I don't believe you can draw any conclusion with the measurement methodology you have chosen.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.