Is there a modern replacement for the HO3508 LA probe?

[Fieroluke]

HI Thomas,

thanks for your input and invaluable research regarding the protocol.

Yes, the input stage took me the longest to reverse engineer, the rest was just datasheet implementations with a few component value variations.

The probe wires are not available from R&S, not even for the new probes. My requests remained unanswered. Their customer service is abysmal, just downloading a PDF requires a registration process that prevents mortals with non-big-company domains from registering. They prefer selling new equipment, along with the wires. If a wire breaks, go and buy a new scope. Wonderful. But I digress.

HP is much more open regarding this, you can purchase these probe leads on ebay real cheap, they look identical to the ones from R&S, so they probably come from the same sweat shop in Bangladesh or the Hunan or Guangdong provinces. I was unable to find a manufacturer in China, so getting HP is the best option.

But - searching for probe wires same as when searching for a mate - it's the inner values that count, and here R&S and HP are different! The chubby part of the probe wire contains a voltage divider including a cap, which determines the input impedance, and here HP uses slightly different values. I think 1:10 vs. 1:9 or something like that. The cap in the probe wire and the corresponding cap in the probe must correlate with the R dividers because using only Rs would result in a divider that's good for DC, but not high frequencies. HP is very open regarding these probe wires, they even include component values to use when including the dividers on the target PCB using debug headers. Way to go HP!

Luckily the whole thought train and calculations were described very well in some application note I found at R&S, so I was able to calculate the correct component values when using HP probe wires instead of R&S wires.

Bottomline: please be patient, when it's all working I'll publish it. Maybe it makes sense to design a tiny flex pcb for just the probe lead divider that you can embed into a small case while 3D printing it (you know, print, pause, put pcb in half-printed case smooth side up, put some goo on smooth flex and continue printing.)

[Fieroluke]

If anyone is interested, page 9 shows details of the Agilent/HP probe tips:

https://xdevs.com/doc/HP_Agilent_Keysight/agilentprobing.pdf

The HP has a 90.9k resistor with likely 9.1K inside the LA (pod in our case), resulting in 100k total, 9.1/100 attenuation.

At low frequencies, the attenuation of the probe is determined by the resistors. At high frequencies, the capacitors become dominant. Then the attenuation of the probe/pod combo depends on the ratio of the capacitors. The capacitance of the pod input capacitor must be selected to form a 100:9.1 voltage divider with the cap in the probe wire. Then the divider works across all frequencies (up to a limit of course). Same as a scope probe btw.

[Fieroluke]

The part number of the probe leads I’m using is 5959-9333 (pack of 5).
Specs can be readily found.

[Fieroluke]

I have implemented a protocol handler emulating the entire protocol described in this thread, and I’m now able to turn the pod on and off using my HMO724. It should emulate the newer pod for newer scopes too.

This is an older scope and it only supports very few messages.

One message is particularly important, as it switches power on and off (all values hex):

83 02 00 00 or 83 02 00 01

I interpret this as Write Variable 3, 2 bytes follow, 16 bit value 0 or 1 for off/on.

Scope expects the response

03 02 00 01: Report(Read) Variable 3, 2 bytes follow and the value that was set.

But when the pod is off, the scope sends

83 03 02 00 01

Again: Write Variable 3, 3(!) bytes follow.
Values 02 00 01.
The response 03 03 02 00 01 is accepted by the scope.

Still, the two variants of the same message are strange.

[Fieroluke]

A brief update in case anyone is watching this space:

- Positive and negative voltage converters are working as designed, 2.56V and -5.2V
- I have calculated the threshold voltage divider for use with HP/Agilent leads, have yet to test it.
- The hardware will work with either HP/Agilent or Hameg/R&S probe leads, but different resistors must be used/omitted so one has to choose.
- I'm waiting for one (hopefully) final part to be delivered, once that's here, I can do final testing and hopefully publish everything.

On a side note:

The R&S ZL-03 isn't such a bad deal, even if it's pricey! The parts alone (ok, ordering from Mouser, but still) run about $75, plus the multilayer board ($25 with shipping). And you need to have fun soldering 0603 and smaller parts, have a steady hand if you use stencil/oven or better yet have the board populated. And purchasing the test leads isn't cheap either, I paid $25 for a pack of 5 (so $50 total), and the ground leads set you back about the same (another $25). Finally the little hooks/test clips, quality ones are also $5 a piece, which comes to another $50 with shipping.

Grand total over $225, if I calculated correctly, and that's without an enclosure or assembly.

So, not a lot of money saved, but a fantastic learning experience...

[Fieroluke]

Minor setback.

A few observations:

The 4 comparators get warmer than I thought. But they’re ECL, so maybe that’s to be expected. I hope.

More seriously, calibration of the pod fails (see image) The manual mentions very little what’s going on except the leads must be plugged in. Since the voltage divider is in those leads, that pulls the inputs to ground. Without leads they’d be floating. Manual says it’s calibrating input thresholds. With all inputs pulled to ground, the comparators will go 0->1 only when the threshold voltage goes negative, right? So maybe the scope steps through the threshold range noting when the inputs toggle. The Z8 also has an input labeled „ANA“, maybe it‘s measuring something? Threshold voltage at the pod maybe?

Either way, the scope is looking for something and I’m missing something…

[Fieroluke]

Success: self calibration runs through now!
Turns out the scope doesn’t send any messages to the probe, it just searches for the threshold voltage for each channel during self calibration until it toggles. Seems part tolerances result in threshold variations. Maybe that’s why the ZL comes with a DAC: to be able to tweak heavy threshold individually…

Did find and fix a small layout flaw though, so scrapping my remaining boards.
I’ll make a final batch of the fixed version though, and I have yet to design an enclosure.

[Fieroluke]

Happy to report development is now complete!
I had to make a few minor changes to the layout, so my spare PCBs are going in the scap.
My pod works with either Hameg/R&$ Probe lead$ or HP/Agilent leads, which are readily available and supposedly cheaper because they’re not hand crafted from pure unobtainium by naked virgin fairies in a magical forest in Narnia..

Depending on which leads are to be used, different resistors/caps need to be used, and since the HP leads are longer, the enclosure must be fitted with a small spacer. But everything is still 3D printable. As a result, input impedance is 100k with the original leads and 113k with HP probe leads. Other than that, identical pod!

I’m going to 3D print the final enclosure on Monday, stay tuned!

[Fieroluke]

Finished designing an enclosure optimized for FDM printing.
I’m very happy with the results…

Now I have to make a nice sticker…

[maxmatteo]

just great stuff,

anyway to get a pcb and a BOM list? would love to use my logic channels on the rtc1002