Hello,
First post here so I hope it's in the most appropriate forum section. I did not find anything useful on a search.
I am trying to measure the rise time of a 10 MHz square wave with a Keysight 53220A universal counter. The source has a specified rise time at 10 MHz of <3 nS. This is confirmed to be about 2.4 nS when measured on a scope.
However, when measured on this counter I am seeing a rise time of around 40 nS (10-90%). Same measurement seen for fall time. All 50 ohm input impedance and both channels show the same results.
I suspect that I may be setting up the measurement incorrectly on the counter. (the user manual has very little on this measurement setup). Perhaps a trigger or gating parameter adjustment needed with the fast edge?
Any help or suggestions would be appreciated.
Thanks!
Welcome to the forum!
Looking through the manual it does appear it can measure rise and fall times. I wasn't able to find the specific minimums it can measure though. But I did want to ask why you wouldn't use the more standard device to measure rise/fall times which is an oscilloscope.
Thanks Steve. Just doing some experimenting with it since the capability is supposed to be there. I wanted to confirm that as well as the accuracy. I was hoping that a forum member might have some experience with this unit and could shed some more light on it. The manufacturers documentation as you can see is somewhat lacking.
Some more testing yesterday did confirm it works on sine waves in the current setup as advertised, at least up to 150 MHz.
The square wave measurements seem to be inaccurate. Perhaps it's a harmonics issue or a timing/triggering glitch due to the faster edge.
I don't see anything in the data sheet relating to rise time either, other than a minimum pulse width of 2 ns. I wonder if the analog front-end had a very sharp roll of beyond 350 MHz that cuts off the higher harmonics. This something you could easily simulate in something like LTspice. But even then I would expect a much faster rise time than 40 ns. Is it a noisy signal (high frequency or jitter) that the triggering system has problems with?
If you were using the 1 MOhm input, I'd suspect signal integrity issues, but with both the scope and the counter set to 50 Ohm that shouldn't be an issue.
What's your trigger system set up? Do you have something like noise rejection enabled? Any manually set trigger levels that are so high that they only rarely get hit?
Fired up an AWG, scope and a 53230A.
Both report similar enough values. I powered on the 53230A, changed to 50 ohm and selected the rise time. I also changed to DC coupling but it made no difference. Mine is a 53230a not a 53220a though, that may make a difference.
Screen shots of the scope and counter attached:
Thanks for checking that for me Steve. Yes, the 53220A is also capable of this function. I believe I have figured out the riddle!
As ridiculous as this may sound. User error...sort of.
Turns out that the unit I am using has the option (201) parallel CH 1 & CH2 rear panel inputs installed.
If one reads the microscopic footnotes in the user manual the front panel inputs remain active and functional however their "performance is not specified".
I contacted Keysight about this and they could give me no details except that there may be "reduced sensitivity at some frequencies" on the front inputs.
Using the front inputs now for about 8 months and I have not noticed any unusual results before but perhaps it was just the type signals I was testing. This is the first time I tried making this specific measurement.
While testing this theory more last night I can see a significant fall in Vpp on the front inputs as frequency rises, especially above 100 MHz. The rear inputs perform perfectly as expected and was able to successfully measure rise times as low as 1.77 nS. (I don't have a source any faster than that to test.)
I was surprised as to the magnitude of this effect. I got the schematic details of this option, and it is simply two cables that plug into the circuit board (MCX connectors) near the front inputs and run them to the rear BNC's. All very peculiar.
As it turns out that in my present bench setup, using the rear terminals is a PITA. For now, I have just connected a 4' cable to rear panel CH1 and ran it forward on my bench for easy access when I need it.
There is clearly a significant difference in certain measurements when using the "specified" rear terminals over the standard front terminals.
Another interesting finding. If I connect a short 2-foot unterminated BNC cable to either front channel connector while using the corresponding rear channel connector it causes the same erroneous reading from the rear channel connector. I think this confirms that the Option 201 add parallel rear connectors, by adding the short length of cable inside the box to run to the rear connectors somehow seems to have a real effect on the front inputs circuit. I wonder if this can be explained by the small added capacitance from the added "stub"?
My question now is if I open up the unit and simply unplug (or completely uninstall the parallel rear connector cables) if that would return the "specified" performance to the front connectors.
Do you think the unit would require some magical re-calibration procedure? I am guessing that's what Keysight would say if they were asked to uninstall it.
Just check the service manual. It seems to contain the instruction on how to install the rear connectors. I didn't look at the details though.
I was given an old HP 53132A counter a few months ago which is the precursor to the 53220A. Whenever I get a counter, I like to hook up a signal generator and crank the amplitude down until it stops counting; this gives a basic check of the front-end sensitivity. Mine also had the rear panel input option but I thought nothing of it and used the more convenient front BNC jacks.
But I soon found that I was losing as much as 20 dB of sensitivity between 100 and 150 MHz. After scratching my head for an afternoon, I checked the datasheet and saw the footnote about the front panel performance degradation when the rear panel input option is present. It is due, of course, to the coax cables running from the mainboard back to the unterminated rear panel jacks. The effect can be mitigated by placing 50 ohm terminators on the rear panel jacks.
Since I had no intention of using the rear panel, I opened up the unit and desoldered the two coax cables. One can leave them inside the unit (taped off to prevent the wires from touching the mainboard), or remove the cables along with the two rear panel jacks.
No re-calibration was required (although I did it anyway because it was simple).
Officially Keysight will state a re-cal is needed if anything is changed internally. If you need to know for sure your unit meets specs then it only makes sense. I am sure you can just unplug the rear connectors and will not notice any difference. Or if you do the difference is very minor. The counter is also very easy to re-cal yourself.