Siglent SDS2000X HD isolated signal generator mod

[TopQuark]

So I was revisiting a wideband injection transformer topic (https://www.eevblog.com/forum/testgear/injection-transformer-for-bode-plot-my-diy-project/), which was created so that you could have a floating voltage source to inject signals into a DUT, often a power supply, to measure its loop response and stability. It then crossed my mind that the internal signal generator of the SDS2000X HD is actually a USB device. As such, it is actually possible to take it out, add an USB isolator between the AWG module and the scope to have an isolated AWG, skipping the need for an injection transformer.

Siglent actually sells an external isolated USB AWG (https://siglentna.com/product/external-arbitrary-waveform-generator-2/) for this purpose, but for some reason the isolation voltage is only rated to ±42 Vpk, and it is some extra thing you need to buy.

So I opened up my scope for the millionth time and extracted the AWG module. I found a type-c breakout board with pinout exactly matching the AWG module footprint, and soldered it to the AWG module. Plugging the AWG module to the front usb port of the scope, it immediately recognized it and worked the same way when it was internal, I then added an USB isolator in between, and it worked perfectly well too. 

[TopQuark]

Comparing this mod to using a wideband injection transformer, the output amplitude and source impedance of the isolated AWG is flat across the specified frequency, whereas the best injection transformer still struggles at the lower and upper frequency ranges (see the post linked above). Depending on the isolator used, the isolation capacitance can also be much lower when compared. My two injection transformer measured 54pF and 228pF, whereas my USB isolator measured 16pF, much lower.

Then I tried using the isolated AWG to measure the loop response of a buck converter (TPS54561EVM-555), comparing the isolated AWG vs signal generator with injection transformer. For the same amplitude profile, the bode plot at low frequency has a much cleaner result, as the AWG does not suffer from core saturation at low frequency, limiting the strength of the injection signal at the low end.

One thing to note, compared to the eval board manual, the bode plot looks quite different at lower frequencies. I suspect this is a goof on TI's side, using a DC load instead of a real resistor as the load. A CCM buck with type 2 compensator should have a pole at origin and have a negative slope at close to DC, by my calculations, the TIA in the converter chip should provide enough DC gain to not have a flat gain until 100Hz.

Anyways, I think I'll design a 3d printed case to house the isolator and AWG in a nice insulated package. This simple hack should prove quite useful for power converter development.

Not sure why my photos are not showing up as expandable thumbnails 

[mawyatt]

Clever

Could this possibly be done inside the DSO with the USB Isolator? Maybe not as the DSO AWG BNC is likely grounded to chassis, altho one might be able to rig up an isolated BNC.

Best

[2N3055]

Siglent SAG1021I is isolated..

[TopQuark]

Siglent SAG1021I is isolated..

... is exactly what I said, and ...

Siglent actually sells an external isolated USB AWG (https://siglentna.com/product/external-arbitrary-waveform-generator-2/) for this purpose, but for some reason the isolation voltage is only rated to ±42 Vpk, and it is some extra thing you need to buy.

why this is still a viable idea.

Could this possibly be done inside the DSO with the USB Isolator? Maybe not as the DSO AWG BNC is likely grounded to chassis, altho one might be able to rig up an isolated BNC.

Thanks Mike  I did find this BNC adapter that would do the job, mechanically it should fit the hole in the case, I ordered one to see if it works. Even with that connector, plastic screws and shims isolating the AWG module from the chassis, the ground plane of the AWG might still couple with the chassis to add some pF of isolation capacitance. I'll see if it works when the adapter arrives.

The other direction I was thinking was an optical fiber isolated USB extender, with a battery on the other end powering the awg, to get "0"pf, many kV isolation, inspired by optically isolated voltage probes. But honestly that is a bit overkill, not sure how that could benefit things.

[mawyatt]

The other direction I was thinking was an optical fiber isolated USB extender, with a battery on the other end powering the awg, to get "0"pf, many kV isolation, inspired by optically isolated voltage probes. But honestly that is a bit overkill, not sure how that could benefit things.

Maybe Wireless USB and a Battery?

Best

[voltsandjolts]

An optical/wireless controlled, battery powered sine generator could be cheaper than the fancy core used in injection transformers