EEZ Bench Box 3 (BB3)

[AlanS]

Well I had another look. 

Module B has a fixed output voltage of about 190mV. When loaded by anything the output voltage drops to zero. When the output voltage setting is increased to 190mV the module goes out of CV mode laving the output voltage at 190mV.

NOW I'll look at the circuits. 

 

[prasimix]

Well I finally got around to building my BB3. What a lovely piece of gear and the basis for a data acquisition unit I need for module testing. Denis, I need the second frame.

I'm full of admiration for Andrew McNamara. He evidently had few issues retrofitting the USB connector. With my old eyes it was no fun at all. The language was colourful and would have been educational for grandkids were they allowed to visit in Danistan - read Victoria, Australia.

I now have one DCP405 module (A) working as expected - the other (B) not so much. Switching AC-DC pre-power supplies and positions in the backplane produces the same results. That is A works and B is different.

B has a maximum output voltage of about 200mV and goes out of CV mode at about this point. Hmmm. Time to look for the circus diagrams.

I'm so sorry to hear that. A couple of suggestions on where to start unless you want to arrange a replacement:
1. see if the PCB needs to be further cleaned of solder balls (some of them was so dirty!)
2. when OE is on, check directly on the gate of Q9 if you have +5 V

I presume that output current is not set to zero when you get about 200 mV.

[Andrew McNamara]

I'm full of admiration for Andrew McNamara. He evidently had few issues retrofitting the USB connector.

It hasn't been refitted yet, unfortunately. I'm waiting on a friend with reflow gear and skills to have time to take a look.

[Kean]

Module B has a fixed output voltage of about 190mV. When loaded by anything the output voltage drops to zero. When the output voltage setting is increased to 190mV the module goes out of CV mode laving the output voltage at 190mV.

Hmmm... that sounds very similar to what I see with my 3 "junkyard" DC405 modules, although I expect there will be several causes for these symptoms.  I haven't dug in too far yet, but two of mine show significant solder balls around the middle of the PCB.  The third looks quite clean with just a few solder balls sitting in tented vias, but that could indicate some problems lurking below... 

It hasn't been refitted yet, unfortunately. I'm waiting on a friend with reflow gear and skills to have time to take a look.

Andrew, I'd be happy to take a look for you - just not this weekend.

Kean

[Kean]

I know my "junkyard" DCP405 boards are not expected to work, so this is just a quick data point during some initial troubleshooting.
All three exhibit a similar problem with only 200mV max output (set for 5V 100mA), but only on one does the CV light stay on, and no CC on the one where CV goes off.
Anyway, all three show a low output voltage from the pre-regulator of about 4.7V instead of roughly double that on a good DCP405.  Not sure if that is a pre-reg problem though.

Gate of Q9 is near 5V when output enabled, and 0V when disabled on all three.  +/-9V and +/-5V rails are all OK as well, although the 9V rails are a bit low at around 8.6V.
Oh, and apparently the one where the CV light stays on apparently doesn't even manage to output 200mV.

I need to study the schematics more and get my head around the design... which was a large reason behind getting these.  Thankfully someone published excellent documentation & design notes!

[prasimix]

Anyway, all three show a low output voltage from the pre-regulator of about 4.7V instead of roughly double that on a good DCP405.  Not sure if that is a pre-reg problem though.

That doesn't necessary means that pre-regulator has a problem because its tracker (Q2) just follows what is on the output. Right way to check pre-regulator is to disconnect R10 from POST_OUT+ and connect it to external control voltage to see if it can output voltage that is few volts above control voltage. It also has to be enabled. If you want to do that externally, remove D1 and apply +5 V on pin 8 (RUN input).

[Kean]

That doesn't necessary means that pre-regulator has a problem because its tracker (Q2) just follows what is on the output.

Yes, I was just coming to that conclusion as I was writing the above.  Getting late here though, and I have to work this weekend... but I'll try get back to it soon.  Thanks!

[Kean]

OK, I fixed one.  Yay!   

The pre-reg was fine, confirmed per instructions from prasimix although I cheated a bit just injecting 5V (from channel 1 with common ground) via 10k into base of Q2.  So I became suspicious of the circuit around Q8/Q9.
Tracing the schematic while checking various things and I discovered R49 was completely missing (15R 0.75W 2010).  It is part of OE circuit alongside Q9.  I only had a 15R 1210 part (0.33W), but installing it gave me an output.
I tested at 50mA load from 2V to 15V and it was OK but uncalibrated.  I need to reinstall the heatsink before I calibrate and run some more tests, but I think it will be good once I also replace that resistor with higher rating.

I don't think the other two will be so easy... but that is OK as I am actually looking forward to studying the control loops further.

[prasimix]

It can be declared functional when you can output all 5 A (for that of course you need to put heatsink back) and DP is functional.

[Kleinstein]

R15 should see a maximum of some 100-200 mA (mainly the base current for the output stage) and thus up to some 150-600 mW. So the smaller resistor may work, if the power transistors are relatively high in amplification. The power at the resistor goes up with the square of the output current, so some 2 A should still be on the safe side with the lower power one.

The control loop may want some more tweak.  This is about the cross over between CC and CV mode. There was trouble at low currents, that could be reduced by turning of the down programmer, but it does not solve all the trouble here.

Missing parts (possibly lost on transport), bad joints and solder balls are likely the prime candidates to look for at the bad boards.

[Kean]

It can be declared functional when you can output all 5 A (for that of course you need to put heatsink back) and DP is functional.

Was trying to go home... but challenge accepted!   

[Kean]

And success!   

Voltage calibration was really easy.  It seems to now be reading within 1mV of set value from 1 to 40V (no load).
Current calibration went ok-ish, but it will need to be redone because some idiot just blew the 10A fuse in their 34465A. Doh! 

I wasn't sure of the best way to test the down programmer... and it is late so I wasn't going to RTFM LOL.
So I set channel 1 to 5V 20mA and connected it in parallel with channel2 set for 4V 1A, and it was happily outputting 4V and sinking 20mA.

[prasimix]

I wasn't sure of the best way to test the down programmer...

The easiest way is to set the trigger mode to list and create a simple list:



Decrease "YT view sampling rate" to 10 (or 5 ms):



Output without load connected and DP enabled:



With DP disabled:

[prasimix]

A brief update on the problem I encountered with the MIO168 module: the ADC ADS8674 used with floated inputs (when nothing is connected to them) will output an internal biasing voltage that depends on the selected range. Let’s say for ± 2.5 × VREF range (i.e. ± 10.24 V) it will be around 2.2 V probably because the leakage current for that range is 2.25 uA and the input impedance is 1 Meg.
It looks ugly and confusing and I don’t know how to solve it without adding an op-amp to the input which I have already discussed with @Kean on Discord.

In Section 8.3.1. that is confirmed, and sounds to me as "It's Not a Bug, It's a Feature":

If the analog input pins (AIN_nP) to the devices are left floating, the output of the ADC corresponds to an internal biasing voltage. The output from the ADC must be considered as invalid if the devices are operated with floating input pins. This condition does not cause any damage to the devices, which are fully functional when a valid input voltage is applied to the pins.

The nice thing about the currently used ADC is that it's multi-range, has a bipolar input and accepts high voltage (± 10.24 V is high from the ADC standpoint) so the AFE is minimal. However, if I have to add op-amps to the input, then I can try my luck with some other ADC and arrange the bipolar input and range as desired. Does anyone have a suggestion for good alternatives?

[Kean]

It looks ugly and confusing and I don’t know how to solve it without adding an op-amp to the input which I have already discussed with @Kean on Discord.

So I had one (ugly) suggestion for this which was to add a 1M load resistor at the ADC inputs which is switchable under software control.
Then if the firmware detects an input voltage in a certain range as typically caused by the ADC PGA bias voltage leakage, the firmware can enable the 1M input impedance and then monitor the ADC reading.  If it drops by approx half, then the input can be considered floating.
This could of course be disabled via a setting if the user wants to avoid that behaviour, or the process could also be reversed with the 1M input impedance normally enabled.

Yeah I know it is an ugly hack.   

[Kleinstein]

Normally it is the responsibility of the user to apply the right signal - to the card does not really have to detect of the input is open. Don't expect a correct measurement if nothing is connected. Adding an external amplifier would not avoid the problem.  There is not real advantage in reading close to zero from an open input.

I think the idea of switching the input to see if the input is not connected is a good idea. However the inputs are differential, so there is more than just the option of open or connected. It can be open at both sides and there may be a relatively high impedance source. To some extend already switching the gain may be enough to detect an open input (so without extra HW).

When used as a true differential input with an isolated voltage source, I see not problem and there should be no offset / extra bias current. The biasing is nice to keep things in the center where the amplifier can work.
The main trouble comes when the ADC is use as single sided - than the 1 M input impedance is not relative to ground, but a fraction of the internal biasing voltage. So one would see some input bias current (0.1-2 µA) and this current in combination with a non zero source impedance can be a problem. It adds some offset and not just the normally expected gain error.

For the price the performance of the ADS8674 still looks good.
With the internal AA filter the 500 kSPS are a bit misleading, as there is a 15 kHz BW limit. At least this is also there when sampling all inputs in a fast sequence.
With external amplifiers/input stages I would consider a different ADC, like ADS131A04 or similar. The sampling rate is lower, but more BW and as a SD ADC not much AA filter is needed.

[Kean]

When used as a true differential input with an isolated voltage source, I see not problem and there should be no offset / extra bias current. The biasing is nice to keep things in the center where the amplifier can work.

Except it can't be used as a true differential input, only single ended.
Per the datasheet:

The devices allow a ±0.1-V range on the AIN_nGND pin for all analog input channels.

[Kleinstein]

When used as a true differential input with an isolated voltage source, I see not problem and there should be no offset / extra bias current. The biasing is nice to keep things in the center where the amplifier can work.

Except it can't be used as a true differential input, only single ended.
Per the datasheet:

The devices allow a ±0.1-V range on the AIN_nGND pin for all analog input channels.

This is indeed a problem. So there should be a way to tie the negative side input with a relatively low impedance  to ground. I would suggest more like 1 K, so that even with 10 V applied to the 1 M input impedance on the high side by an external source would not drive the return side to more than 100 mV. It is still nice to have some kind of low side sensing for the resistors used for current inputs.

[Andrew McNamara]

It hasn't been refitted yet, unfortunately. I'm waiting on a friend with reflow gear and skills to have time to take a look.

Andrew, I'd be happy to take a look for you - just not this weekend.

Thank you for the offer Kean, but my friend was able to successfully reattach the USB connector tonight and I've since updated to v1.4 (very nice!)

[prasimix]

I'm working of "finalization" of MIO168 module. As you probably know it should offer mixed I/O (hence MIO) where both digital and analog signals can be processed. The following idea came to my mind: as we already have a "power bus" on BP3C for different types of power output coupling, why not add an "analog bus" through which certain signals can be passed without the need for external cabling. A new bus would be placed on top of the module, with e.g. 2mm IDC connectors.

I wonder if someone may have already come across this way of connecting to refer me to a brand or model that has it, so I can see what they have to offer on such a bus.

[danielbriggs]

I'm sorry, not suggestions / info on the bus examples for other T&M, but for the physical implementation: what about PCB edge connector + something like a Nvidia SLI bridge?

They do 2x and 3x way ones, and the module spacing looks similar:

[prasimix]

Looks expensive

[Kleinstein]

Some analog interconnection can make sense, e.g. from an analog input or output to a relay card.  It may include also some digital (e.g. sync signals).
It could still be tricky to find a good layout / choice of pins to make it work for more than a few cases that are know up front.
Additional switching may be needed on the cards, if one wants to keep the flexibility without changing the interconnects.   

The IDC connectors look OK. I would probably go with the more standard 2.54 mm pin spacing and this 1.27 mm spacing of the normal flat cables. I would prefer more like a low pin count (e.g. 10 or 6 pins) and if really needed to use 2 cables. This may give more flexibility, to choose only the wanted connections.

[jbb]

I guess the ‘obvious’ application is to connect signal relay MUX cards to a hypothetical DMM card.

For that you actually wouldn’t need many wires (e.g. a ‘deluxe’ implementation of ground, guard+, force+, guard+, sense+, guard+, guard-, force-, guard-, ground) so 8-10 pins would probably work. Of course, if you might want to move some current around, you’d need paralleling.

I think Kleinsten’s right about using standard 2.54mm pitch and leaving room for different jumper options. If you go 15-way could you reuse parts from the main board to auxiliary board IDC?

On ribbon cable vs rigid PCB: cables have their issues but are tolerant of misalignment. A previous design I did had a PCB base board on the bottom and a PCB bridge board on the top and it got quite hard to get all of my cards lined up and plugged in without one leaving over.

Of course, it might be better to revisit the backplane for v2

[prasimix]

Thanks @Kleinstein, @jbb. I like idea to split analog bus connector in two, but don't like idea to use 0.1" pitch since my current experience with low pin count cables is that have a weak connection, i.e. they easily lose contact and for that reason I would go for a smaller pitch.