Output load for testing

[morls]

What is the best way to apply a load for testing frequency response? I'm looking for a load of 100K, and another of 10K or more, so I suppose 100K would also work in this instance.

I'm using the built-in signal generator on my scope, probing amp/mixer input and output to generate Bode plots. I thought I might connect the output to my TC Reverb 4000, which has an input impedance of 20K. If I connect the probe to the mixer output XLR pins, with the cable going to the TC, would this do the trick?

Cheers
Stephen

 

[JohnRoberts]

For testing frequency response of line level signals we generally use a "bridging" termination. Bridging means that the output or source impedance sending the measurement signal is 1/10 or less of the input impedance of the stage receiving the signal to be measured.

JR

 

[ccaudle]

What is the best way to apply a load for testing frequency response? I'm looking for a load of 100K, and another of 10K or more, so I suppose 100K would also work in this instance.

I'm using the built-in signal generator on my scope, probing amp/mixer input and output to generate Bode plots. I thought I might connect the output to my TC Reverb 4000, which has an input impedance of 20K. If I connect the probe to the mixer output XLR pins, with the cable going to the TC, would this do the trick?

Cheers
Stephen

Connecting another device as load as you suggest can work. Make sure the load device is powered on, in case the input impedance changes when it is powered off.

What kind of probe are you using, just the usual single-ended 'scope probe? If you are doing this as a general technique, make sure you understand what type of output you are measuring, a transformer coupled output will behave differently than an op-amp based output, necessitating a different point of probe ground connection.

Also, if you will be doing this very often then some connectors with load resistors soldered in are very useful.

 

[RuudNL]

Most modern equipment would have an input impedance of 10 or 20 K.
Very seldom 100 K.

 

[morls]

What kind of probe are you using, just the usual single-ended 'scope probe? If you are doing this as a general technique, make sure you understand what type of output you are measuring, a transformer coupled output will behave differently than an op-amp based output, necessitating a different point of probe ground connection.

I'm using a single-ended scope probe, and generally measuring transformer coupled outputs. The signal generator output is via BNC, so I'm treating this as an unbalanced signal in, with pin 3 grounded both into and out of the DUT. The probes are connected to pin 2, with the ground clip connected to pin 3.

Also, if you will be doing this very often then some connectors with load resistors soldered in are very useful.

Would this be as simple as inserting a resistor between pin 2 and probe? So, for a 10K load, a 10K resistor?

Most modern equipment would have an input impedance of 10 or 20 K.
Very seldom 100 K.

At the moment I'm testing a RND summing mixer, the Satellite 5059. I bought it brand new, but there was an issue with one of the outputs and the transformer was replaced. The specs given are:

100K Load
+/- 0.1dB from 20Hz to 20kHz
-3dB @ 3.5Hz / 185kHz

Here's a Bode plot with a 10K load on the output, testing setup as above:

 

[warpie]

I'm using a single-ended scope probe, and generally measuring transformer coupled outputs. The signal generator output is via BNC, so I'm treating this as an unbalanced signal in, with pin 3 grounded both into and out of the DUT. The probes are connected to pin 2, with the ground clip connected to pin 3.

Would this be as simple as inserting a resistor between pin 2 and probe? So, for a 10K load, a 10K resistor?

At the moment I'm testing a RND summing mixer, the Satellite 5059. I bought it brand new, but there was an issue with one of the outputs and the transformer was replaced. The specs given are:

100K Load
+/- 0.1dB from 20Hz to 20kHz
-3dB @ 3.5Hz / 185kHz

Here's a Bode plot with a 10K load on the output, testing setup as above:

View attachment 87034

I wonder whether the ringing is because of lack/ wrong zobel network or because of wrong test setup. BTW, what keysight is this? Does it have a built-in gen? All in all are you happy with the scope? Sorry for the OT but i'm looking for a new scope

 

[morls]

I wonder whether the ringing is because of lack/ wrong zobel network or because of wrong test setup. BTW, what keysight is this? Does it have a built-in gen? All in all are you happy with the scope? Sorry for the OT but i'm looking for a new scope

I hope it's an error in my setup, I have very little experience with this type of testing.

The scope is an EDUX 1052G, the G signifies a built-in generator. Early days, but I really like this scope.

 

[ccaudle]

I'm using a single-ended scope probe, and generally measuring transformer coupled outputs. The signal generator output is via BNC, so I'm treating this as an unbalanced signal in, with pin 3 grounded both into and out of the DUT. The probes are connected to pin 2, with the ground clip connected to pin 3.

Good, you already have covered my question about measuring across a transformer correctly.

Would this be as simple as inserting a resistor between pin 2 and probe? So, for a 10K load, a 10K resistor?

The resistor would go between pin 2 and pin 3. Depending on how many adapters you wanted to build, I would probably make some adapters with XLR female connectors with a resistor between pins 2 and 3, and a short coax stub that connected the center conductor to pin 2, the shield to pin 3, and had a BNC on the other end, with a label on the connector with the resistance value and something like "floating output only" or "transformer outputs only" to remind myself that the adapter has pin 2 connected to scope ground.
For solid state outputs I would probably do the same, but connect the coax shield to pin 1. You could also split the resistance value in two and have two resistors of half value going from pin 2 to pin 1 and pin 3 to pin 1 if you wanted to emulate a non-transformer input stage, but in principle it shouldn't matter.

 

[morls]

The resistor would go between pin 2 and pin 3. Depending on how many adapters you wanted to build, I would probably make some adapters with XLR female connectors with a resistor between pins 2 and 3, and a short coax stub that connected the center conductor to pin 2, the shield to pin 3, and had a BNC on the other end, with a label on the connector with the resistance value and something like "floating output only" or "transformer outputs only" to remind myself that the adapter has pin 2 connected to scope ground.
For solid state outputs I would probably do the same, but connect the coax shield to pin 1. You could also split the resistance value in two and have two resistors of half value going from pin 2 to pin 1 and pin 3 to pin 1 if you wanted to emulate a non-transformer input stage, but in principle it shouldn't matter.

Thanks, that gives me a clear idea of what I need.

All my gear is hardwired via (balanced) connections to longframe patchbays. The Satellite 5059 inputs are DB-25 connections. I'm in the process of making up longframe to XLR cables so I can connect preamps/AD to patchbay outs. I'm also doing the testing via the patchbay, the test setup is

Signal Generator ⇒ patchbay input via adapter with probe 1 connection⇒ DUT ⇒ patchbay output to adapter with loading and probe 2 connection

I don't know if having the patchbay between DUT and scope would affect the measurements...

 

[ccaudle]

I don't know if having the patchbay between DUT and scope would affect the measurements...

It should only add a small amount of resistance and capacitance to the connection. If you see something unexpected maybe you would want to go directly to the device input and output to isolate whether the patchbay connections were causing any change, but if the device wasn't capable of driving several feet of cable and a couple of connectors you would want to know that.
If this is just for your own personal equipment then test it how you use it. If you are testing equipment for other people, then some adapters to connect up some worst case conditions might be warranted. Slightly different style of equipment, but one lab I used to work in had a big spool of cable with a connector on both ends so we could test gear driving several hundred feet of cable. We could have possibly done the same thing with short cable and attenuators, but with a real spool of cable we didn't have to worry about whether the lab setup accurately mimicked the real world use, it was a pretty straight forward case of spec says this should drive x feet of cable within these performance limits, here is a spool of x feet of cable, connect it up and test.

 

[morls]

Thanks, great info.

 

[warpie]

I've been thinking of making some XLR-to-BNC cables for quite sometime now but haven't got the chance yet.

Am I right to believe that the following cable is quite useless since there's no shielding whatsoever? Not only for xlr-bnc but for any kind of work.

 

[trobbins]

Are the input and output grounds of your mixer/amp isolated from each other (and from mains power) or connected through the device?

 

[ccaudle]

I've been thinking of making some XLR-to-BNC cables for quite sometime now but haven't got the chance yet.

Am I right to believe that the following cable is quite useless since there's no shielding whatsoever? Not only for xlr-bnc but for any kind of work.

Hard to tell just from the picture, but my guess would be that the cable is standard coax up to the molded strain relief near the end, and the coax is joined to the red and black pigtails at that point.

And non-shielded cable is not completely useless, twisted pair cable with tight twists can reject a lot of magnetically induced interference, so depending on the circuit impedance and the frequency range you are investigating unshielded might be fine.

 

[morls]

Are the input and output grounds of your mixer/amp isolated from each other (and from mains power) or connected through the device?

The grounds are connected to pin 3/chassis at input and output. The output is transformer coupled. I'm pretty sure the mains ground would be separate from the audio ground.