Frequency Response Testing with Transformer balanced equipment

[Bingo]

Hello Legion of QuantAsylum’ers,

I started using a QA401 for the first time to test a device… mainly just Frequency Response plugin’s ( regular and Chirp).

I keep getting a huge HF rise starting about 1k ( attached screenshot)… like crazy rise…

I thought something was wrong with the DUT…

But, just to be sure, I disconnected and did loopback from OUT to INPUT of QA401, and got perfectly flat response…mmmmmmm…guess my DUT is damaged.

But, continued testing…

1. an electronically balanced AD/DA - this gave completely flat response just like the loopback…
   mmmmmm… DEFINITELY! my DUT is bad!

But, I tested one more device… one I know is good- but is similar somewhat in design.

2. Transformer Balanced line amp - this unit gave almost the exact same Frequency response results
   … same huge rise in HF starting at 1k… ODD!!.. Both devices broken?.. probably not…

So, I started trying different connections to DU… added load resistors( even though I know both devices are properly terminated… but why not try it,… same RF rise when running Frequency Response Plugins…

Went on to try different cabling… shorting ( -) to ground on cables connecting I/O of DUT… still same
test results.

So, humbly, I come before the council of Quant and beg for enlightenment… why do transformer balanced DUT behave this way?.. what am I missing?.. I have datasheets for the devices and know that they are essentially flat from 20 to 20k …

Very curioius to me that both devices would have this same result.

Very much appreciate your thoughts and help on this.

Thank you,

Bingo!

[Bingo]

P.S. my settings and also test results for DUT #2 ( which is actively balanced- no transformers)

[Bingo]

[matt]

Hi @Bingo, remember the audio analyzer has some series output R. The QA401 has 47 ohms, and the QA40x has 100 ohms. And the input has 100K. The expectation for transformers is that they are matched to their source and load impedances. Usually, this means a series R is added to the inputs and a parallel R is added to the output.

Can you share the datasheet for the these transformers? Additionally, if the manufacturer has provided a spice model, you can run the xfmr in spice with the default in/out impedances and see how it compares to expectations. And then play withe the surround resistors to tame the response curve.

I know you said they were property terminated, so may it’s a source impedance issue?

[Bingo]

@matt thank you so much for the quick and detailed reply to my post. : )

The input to the DUT is 600 ohm and output is 600 ohm as well.

Schematic is here:

By the way, the 2nd device I tested has 20k input ( transformer balanced), 600 ohm output…

circuit much the same as link.

I did put a 620 ohm resistor across the output of the DUT ( from + to - ) on secondary of output transformer to simulate the load it expected to connect to ( since I expected the QA401 to have a high impedance( did not know it was 100k ! - thanks again!)

Very much looking forward to your thoughts on this one.

Bingo!

[Bingo]

[1audio]

It looks like a ground return is missing somewhere. You seem to be getting capacitive coupling from the transformer windings. Are you connecting to the QA401 differentially? If not is the unused input shunted with a termination? And the transformer return connected to ground? Remember that the transformer secondary is fully isolated (the reason for having the transformer).

[Bingo]

@1audio -thank you for your response : )

Very interesting! I thought I had the device connected correctly… but I am attaching a photo of the

wiring I used to connect the DUT and the QA401. Basically, I have 2 cables that go from XLR to dual

RCA Cable 1 is XLR-M to dual RCA cables… Cable 2 is XLR-F to dual RCA .

I then use RCA-F to BNC adapters to connect to the i/o of the QA401.

I have attached a drawing of the connects to detail more…

One more thing, the DUT chassis is on PIN 1 of the XLR which I connect to both of the RCA cables shroud… so I believe that means the DUT’s Chassis connects via the cable through the BNC adapter to the ground of the QA401(?)

The DUT’s chassis is connected mains earth and, to the electronic earth of the circuit ( only at one point).

Thanks in advance-really looking forward to your thoughts and insight ion this!!!

Bingo!

[1audio]

I think you will need to ring out the cables and their connections to the DUT. Ideally a cable with 2 RCA’s to an XLR will have the center conductors connected to pins 2 and 3 and the shields to pin 1. I would then use an Ohmmeter on the cable connected to the output side and check the DC resistance between the two RCA center pins with the cable attached. You should see the transformers DCR. Then do similar with the input cable to verify that the connections are going all the way through to the opamps. The input would need the ground reference from the QA for the OPAMPs to see the correct signal.

Deaing with balanced transformer isolated interfaces is alway a bit confusing. There are pleanty of ways for them to not be connected correctly.

If all these are correct I would check the primary of the transformer and its connections to the PCB for opens.

[Bingo]

@1audio - thank you so much for that info!

I tested the DCR between shield and + and shield and - for both Y cables and found that for the

XLR-M to BNC cable the DCR ( when connected to the QA401 ) was different:

DCR for + to shield = 85 ohms
DCR for - to sheild = 10 ohms

… swapping cables the difference between stayed the same ( even though

I had the cables connected to QA401 opposite original configuration( to me, proving the difference in

resistance was in the cable- not in the QA401- HUGE RELIEF

Tested the XLR-F to BNC cables ( which connect to QA401 “INPUT”, and

there was no measurable difference between + and shield and - and shield when connected.

… actually, the DCR was extremely high- my meter gave up at 25Meg.

I am going to test cable number one… clean up solder flux, look under magnification for possible

sources of capacitance between conductors.

Keep you posted!!

Bingo!

[1audio]

With the cables disconnected you should have open circuit between the shields and + or -. However you need to check that the center pin on the XLR’s connect to either pin 2 or pin 3 with a very low resistance. Were these premade cables? There are several different “standards” for those cables. Some connect the shield to pin2 and the signal to pin 3 and no connection to pin 1.

The input to the QA should measure 100K to ground. If you get something else there may be a short in the cable.
This is how the cables should be wired.:
XLR to two RCAs
1 ---- shield both cables
2 ----- center pin one cable
3 ----- center pin other cable.

[1audio]

If the QA input is cap isolated then you would get an open circuit at DC on the input connections.

[Bingo]

@1audio

Hello, well, I thought I had the issue figured out when I measured the cables between center pin 1 and shield and center pin 2 and shield… there was a significant difference between them.

1. DCR between center pin 1 and shield = 12 ohms
2. DCR between center pin 2 and shield = 85 ohms

There was a little PCB in the middle of the cables( I bought them off Amazon), so I thought I had found my culprit… I removed the PCB and hard wired all connections- exactly as you indicated.

However, when I reconnected to QA401, the same DCR measurements were happening… Wait! I thought, perhaps it is the RCA to BNC adapters!.. swapped them between cables, but the issue did not follow the adapters( DCR measurments stayed the same…

So, I decided to remove the cables and test the DCR between the center pins of BNC on QA401 and shield of QA 401.

I still got the same DCR measurements

Left + pin to shield =12 ohm
Left - pin to shield= 85 ohm

Ahah!!! the QA is at fault!!!

But it gets weirder,… I measured the Right channel of the output… and it gets almost exact same

DCR measurements between pins and sheild…

And again, weirder,… I decided to remove the USB cable from the QA401 and then re-measure DCR
from center pin of OUTPUT BNC and sheild… and got these measurements:

Left

• =110k ohm between center of BNC and sheild

• =109.3k ohm " "

Right ( DCR between center pin of BNC and sheild)
+= 110.2k
-= 109.7k

So, connecting the device to the PC acutally changes the DCR on output between pin and sheild

I decided to do a reset on all settings in QU Analyzer 1.925 … but no help.

Swapped Max INput llevel between +6 and +26 … and no help

This thing is really cute and small… but it is tribbling me!!!

WTT?!!!

I was really excited to use the QA, but it seems Really, Really wonky and at this point looks

to be defective… or a bug with software.

Going to try different USB port or re-install QA software… but getting really discouraged.

Quant Asylum… are you out there?

Bingo!!

[Bingo]

swapped out USB Cables, usb ports… also, uninstalled software and went back a few versions on the Quant Asylum software.

Uninstalled re-installed drivers, rebooted PC still no help.

[Claudio]

@Bingo

Thanks in advance-really looking forward to your thoughts and insight ion this!!!

Hi, Just an observation, which probably has nothing to do with it. From the drawing it would appear that you have connected the 50-ohm terminators on the QA401 generator outputs as well. The unused QA401 generator outputs should be left open, not terminated. Only the unused inputs should be terminated. But maybe I am misinterpreting your drawing.

[Bingo]

@Claudio

Thank you very much for that information. I did have terminators on the output terminals of the channel R channel ( that I am not using).

Do you think this would affect results on the L channel while testing only a mono device using the R channel.

Bingo!

[Claudio]

@Bingo

Do you think this would affect results on the L channel while testing only a mono device using the R channel.

Hi. I don’t know, but I think not. You could try removing the 50 Ohm terminators and see what happens (by the way, unused outputs should be left open, only unused inputs should be terminated, again as a fixed rule).

[matt]

Hi @Bingo,

So, I decided to remove the cables and test the DCR between the center pins of BNC on QA401 and shield of QA 401.

If you plug in your QA401 and remove all the cables, and then measure R across the outputs with your DVM, you should see about 47 ohms on all outputs. That is, connect one side of your DVM to the center pin of L+ and and the other side of your DVM to the BNC shell, and confirm you read about 47 ohms on all four output BNC.

For the inputs, the DVM won’t be much help here because the inputs are AC coupled.

Can you confirm your outputs are all about 47 ohms with the QA401 plugged in (software doesn’t need to be running). if you are really measuring 110k or so, then the output resistors have opened and will probably need to be replaced. You can use a 0.1% 47 ohm 0603.

As @Claudio notes, it’s not good to ask opamps to drive into short circuits. When you place shorting blocks on the outputs, you are asking the output opamps to drive into much lower loads than they are expecting to see.

[Bingo]

Hi @matt ,

Thank you so much for this information.

I have tested the device previously with the process you described:

“pin of L+ and and the other side of your DVM to the BNC shell, and confirm you read about 47 ohms on all four output BNC.”

However, my results were that, if USB is plugged in, I get ~12 ohm when testing DCR for + to BNC shield( on either L or R ) channel-

and get ~85 ohm when doing the same measurement between - and bnc shield.

odd to me that the measurements were similar between L and R output channels… but that

• to shield was different than - to sheild.

Bingo!

[Bingo]

Matt,

Thank you so very much for this information.

I have measured the OUTPUT of the QA401 and attached a photo of diagram with the measurements.

next to each BNC connector… ( in parenthesis are the measurements for same BNC when QA401 is not plugged into USB).

I did notice one odd thing, that when measuring the same ports ( an flipping the probes polarity, the test results are different… measuring essentially the inverse test results…for instance, if I measure
DCR between center of BNCL + and sheild it comes back at 81.9 ohm… but if I flip the probes- now black probe goes to center pin and red probe to BNC sheild- the test results are 13.3 ohm…
And, for L - BNC, i get initial test result of 10.1 ohm… ( flip probes ) and DCR is 84.2 ohm

Anyway, I’ll open up and test where you indicated…now that you mention the shorting blocks, If that is the case, then that could explain the R + and R- being wonked( as I did have them on those terminals) but, I have never had the 50 ohm terminators on the L+ or L- outputs.

Will report back as soon as I tested the 4 resistors you indicated.

Thanks again!

Bingo!