-
OK, so I've been working on documenting my test setup for the verification of the design specifications.
I've already placed some test results of these tests here, and plan to release all orderly on Github.
It's a total of 12 measurements with their own setup. I've the text here, with pictures of the setup, or a HTML collated.
I'm open to discuss the test setups, let me know if you think I'm doing something wrong or you know of a better way.
Text is in markdown, the HTML is nicer, but not supported as attachment, so I have it as PDF.Code: [Select]# Measurements done on the PRP1 Power rail probe
This documents lists all the measurements that are done on the PRP1 for design characterisation.\
The measurements were done on TBD (at least two) separate units, results are published in folders
## 1) Bode plot 10MHz
Verifies the gain errors across the frequency span\
\
Equipment used:
* Siglent SDS804X HD Oscilloscope
* Siglent SDG1032X HD Generator
Settings:
* 10Hz to 10MHz
* 121 point
* 100mV peak to peak
* 0V offset
* Frequency mode: Linear
* DUT input CH1
* DUT output CH3
* Load 50 Ohm
Pass criteria: 1dB maximum deviation between minimum and maximum of the gain plot
Note: Crossover frequency is ~ 100 KHz\
Note: 1.3:1 is 0.77 or -2dB
## 2) AC gain @ 10 MHz
Verifies the 1.3:1 probe attenuation\
\
Equipment used:
* Siglent SDS804X HD Oscilloscope
* Siglent SDG1032X HD Generator
Settings on generator:
* 10MHz sine wave
* 100mV peak to peak
* 0V offset
* Load 50 Ohm
Pass criteria: 5% maximum deviation between RMS and Peak to peak value of CH1 and CH3\
## 3) Noise measurement 20MHz BW 12 bit Scope
Verifies the noise figure of the PRP1\
\
Settings:
* PRP1 zeroed-out
* PRP1 powered by lab linear power supply
* PRP1 input shorted by SMA shoring plug or 50Ohm load
* 2 mV/div on CH1 and CH3
* 1 ms/div time scale
* 20MHz bandwidth limit
Pass criteria:
$\ Noise = sqrt{CH3^2 - CH1^2) < 50uV$
## 4) DC gain
Verifies the 1.3:1 probe attenuation on DC\
\
Equipment used:
* Siglent SDS804X HD Oscilloscope
* Siglent SDG1032X HD Generator
Settings:
* PRP1 zeroed out on Oscilloscope with 0V input
* 0.1 Hz square wave
* 100mV peak to peak
* 0V offset
Pass criteria:
Measured 100mV peak to peak on the oscilloscope on CH1 and CH3\
Two measurements are 5% within each other
## 5) Cancellation range +24V and -24V
Verifies the +/-24V Cancellation range of the PRP1\
\
Equipment used:
* Siglent SDS804X HD Oscilloscope
* Rigol DP832 Power supply
Settings:
* PRP1 zeroed-out after each change of the voltage on power supply
* Set up 5V, 12V, 24V with 0.1A current limit on power supply
* Flip BNC to banana adapter and repeat measurement
Pass criteria:
PRP1 can be adjusted to less than 100mV measured on the oscilloscope
Output current of the power supply is less than 1mA
## 6) 500mV dynamic range
Verifies the active signal range of the PRP1\
\
Equipment used:
* Siglent SDS804X HD Oscilloscope
* Siglent SDG1032X HD Generator
Settings:
* PRP1 zeroed out on Oscilloscope with 0V input
* 0.1 Hz square wave
* 500mV peak to peak
* 0V offset
Pass criteria:
Measured 500mV peak to peak on the oscilloscope
## 7) DC input 50KOhm
Verifies the 50KOhm input impedance in DC conditions of the PRP1\
\
Equipment used:
* Siglent SDS804X HD Oscilloscope
* HP 34401A multimeter
Settings:
* PRP1 zeroed out on Oscilloscope with 0V input
* resistance measurement auto range
Pass criteria:
50Kohm +/- 2% measured on DMM
## 8)a VNA bandwidth 10MHz-3 GHz
Verifies the minimum 2 GHz bandwidth of the PRP1\
\
Equipment used:
* LiteVNA
* Rigol DP832 Power supply
* MegiQ VNA calibration kit
* LiteVNA SS405 Coax cables
Settings:
* 10 MHz to 3 GHz sweep
Pass criteria: 1dB maximum deviation between minimum and maximum of S12 up to 2 GHz\
Nominal gain is -2dB +/- 1dB\
Note: 1.3:1 is 0.77 or -2dB
## 8)b VNA bandwidth 400MHz-3 GHz 2nd setup
Verifies the minimum 2 GHz bandwidth of the PRP1 with a more accurate VNA\
\
Note: Minimum frequency setting of the VNA is 400 MHz\
Equipment used:
* MegiQ VNA-0440e
* Rigol DP832 Power supply
* MegiQ VNA calibration kit
* LiteVNA SS405 Coax cables
Settings:
* 400 MHz to 3 GHz sweep
* S11,S12 measured
Pass criteria: 1dB maximum deviation between minimum and maximum of S12 up to 2 GHz\
Nominal gain is -2dB +/- 1dB
## 9) S parameters
\
Equipment used:
* LiteVNA
* Rigol DP832 Power supply
* MegiQ VNA calibration kit
* LiteVNA SS405 Coax cables
Settings:
* 10 MHz to 3 GHz sweep
Pass criteria: None
S parameters provided for reference
## 10) 50 Ohm loading at high freqency, VSWR
\
Equipment used:
* MegiQ VNA-0440e
* Rigol DP832 Power supply
* MegiQ VNA calibration kit
* LiteVNA SS405 Coax cables
Pass criteria: VSWR < 1.5 up to 2 GHz\
This is equivalent to Return loss < -13dB\
Or capacitive loading of ~ 0.5pF
## 11) Noise measurement 1GHz BW 8 bit scope
Verifies the high freqency noise content of the PRP1\
\
2 GHz scope should be used for ideal conditions, but I don't have access to one\
Equipment used:
* Keysight MSOX 3104T Oscilloscope
* Keithley 2221A Power supply
Settings:
* PRP1 zeroed-out
* PRP1 powered by lab linear power supply
* PRP1 input shorted by SMA shoring plug or 50Ohm load
* 2 mV/div on CH1 and CH3
* 1 ms/div time scale
* 50 Ohm input on scope
Pass criteria:
TBD
## 12) Power consumption
\
Equipment used:
* Siglent SDS804X HD Oscilloscope
* Rigol DP832 Power supply
Settings:
* 5V, 0.1A
* PRP1 zeroed-out
Pass criteria:
Power consumption less than 20mA
Managed to write frequency wrong 3 different ways.
-
https://github.com/tszaboo/PRP1
I've placed the following on Github:
User manual
Design verification test results
Design verification updated guide
Important change that I increased the specification for the noise. I originally measured the noise with 1:1 setting and that's technically not correct, since it's a 1.3:1 probe, therefore the new specification is 65uV 20MHz.
For me, this concludes the design verification phase.
I'll continue to assemble the PRP1s, test them, and to order the accessory cables. One cable I also need to make myself.
Please see below an extra oscilloscope shot I made during the verification. It's noise coming from an USB power supply CH1 measuring it with regular oscilloscope probe, CH3 with the PRP1 with the 2.54mm to SMA cable. We can see how significantly less noisy the PRP1 is compared to a standard probe.
-
https://github.com/tszaboo/PRP1
I've placed the following on Github:
User manual
Design verification test results
Design verification updated guide
Important change that I increased the specification for the noise. I originally measured the noise with 1:1 setting and that's technically not correct, since it's a 1.3:1 probe, therefore the new specification is 65uV 20MHz.
For me, this concludes the design verification phase.
I'll continue to assemble the PRP1s, test them, and to order the accessory cables. One cable I also need to make myself.
Please see below an extra oscilloscope shot I made during the verification. It's noise coming from an USB power supply CH1 measuring it with regular oscilloscope probe, CH3 with the PRP1 with the 2.54mm to SMA cable. We can see how significantly less noisy the PRP1 is compared to a standard probe.
Just wanted to say, nice work !!
Results look quite good. -
I had some time this weekend to test and put together the first batch.
I'm still waiting for the cables and a sample of the shipping box.
I expect I can call these ready in about 3 weeks time.
-
What would it cost?
Since reading this article, I have become very interested in it:
https://www.siglenteu.com/power-integrity-measurement-solution/ -
What would it cost?
250 EUR + Paypal and shipping fees.
Since reading this article, I have become very interested in it:
https://www.siglenteu.com/power-integrity-measurement-solution/
Siglent is very right about those measurements. Switching noise can and will couple into passive probes. You are trying to measure ripple on a power supply rail, but those spikes make it impossible, since they superimpose and make a few millivolt ripple into hundreds of millivolts of noise spikes. Often times it's so short and it lines up with your trigger, you don't even see it on the scope screen. Plus, thats not what's on your board, it's something that couples into your 1MOhm/10MOhm passive probe as EMI.
Now, power rail probes that are made for your scope are great, and if you (your company) can afford one, buy that. It shows the DC component, tells you if the probe is in overload. Much better for reporting and design verification. And they have cable compensation, or DC calibration. But they are very expensive. The PRP1 won't show you the DC offset on the screen, or tell you if it's in overload. So you have to be aware of the limitations. On the other hand, you can see from the scope screengrabs that I posted, it's less noisy than a x10 probe, and doesn't show the EMI switching noise. -
I've finally had time to finish the remaining test, the second VNA measurements with a higher quality VNA (Megiq). I could do a sweep between 400MHz and 4GHz on 3600 points, so every MHz.
S11, SWR and others are just informative, this is not a pure 50 Ohm system since the source impedance is 0 Ohm and the load is 50 Ohm, and I have parts in series. But it's still tells a bit about the system performance, so I included it.
More important is the gain VS frequency. Nominal gain is 1.3:1 or -2dB. Normally scope manufacturers define the bandwidth of a probe or a scope at the -3dB point compared to the nominal gain. The -1 dB point is 2.7GHz for this probe, and the -3dB is over 4GHz, meaning that the probe fulfills the 2GHz design requirement. I doubt that there are many people with 2GHz+
scopes that would be interested in buying this.
The choice of cabling and the scope's internal 50 Ohm (or external 50 Ohm) will be the limiting factor in all cases.
-
Cables arrived, packaging sample arrived, and it's a nice one.
I only have to order a bit more from the packaging and I'll be ready to ship out the first batch.
-
Looks really good
-
So after some family stuff and then some sickness, I'm ready to continue.
First batch is almost ready. I can send them out this weekend maybe early next week.
I'm accepting orders for the first batch now. PM me.
Price is 250 EUR + payment and shipping fee. Shipping is calculated as DHL smallest package from Netherlands. -
I received my PRP1 today. It will be a few days at least before I can spend time with it, but it looks great.
A few pics are attached in case anyone is curious, but I have not taken it apart... yet...
-
Hi, some update. I've sent the first few PRP1s to people who ordered it. I have just 3 of the batch left.
If I decide to do another batch that's likely to only happen in 2025, and probably going to be a group buy, so it could be a while. -
Mine arrived the other day and I had a quick moment to get familiar with it today by doing some quick measurements.
In short, does what I expected so far and matches the described specifications well enough. I'm looking forward to using it seriously, and happy to add it to the toolbox.
Overall a solid first-batch of a hobby(?) project. Throughout the process (questions prior to purchase, ordering/shipping) tszaboo was helpful and responsive!
I'll come back after I've used it enough with real projects to actually provide valid feedback.
Tried to match test setup against tszaboo's published setups as per GitHub doc
Screenshots attached are named accordingly.
DC Resistance - 49.9829 kΩ when zeroed, 4-wire measurement on 2701.
Power consumption - 5.004 V from DP832. 13.201116 mA measured with 2701 (filtered, 5 PLC)- All tests were done with a lab PSU providing 5V rather than scope USB.
- My scope doesn't support 1.2 input attenuation. Both channels are left at 1x, 50 Ω internal, no BW limit unless mentioned.
- The PRP1 is always on CH1.
- Used my own BNC and SMA cables, adapters etc. All first owner+name RF brands etc, all overkill for >6 GHz work.
- I noticed during the gain test that my sig-gen (DG1022) wasn't too great at hitting the Vpp setpoints. The output BNC was into a tee then connected to the PRP1 and CH2 as reference. I'd expect relative comparisons to be OK.
- I'll try to setup the RSA to look at spectral performance and attempt crude replications of the VNA measurements
There are some screenshots of a quick real-world setup looking at the output of a 12 V buck-boost reg built on the TI TPS552892.
It's running in step-down PWM mode with a 2 A load, and I've intentionally removed the snubbers on the switching nodes.- The SMA pigtail included with the PRP1 kit is connected (poorly) to PTH testpoints on the output.
- A 10 MΩ passive probe (CH2) using the default ground antenna/clip is poking at the same output test-point.
Because the rail is around 12 VDC, CH2 hits the 10 V vertical offset limit and can't be used with DC coupling and a sensitive scale at the same time.
The coarse+fine offset of the PRP1 handles this as expected, while showing a better representation of the signal.
The screenshot can't show it, but the blue CH2 trace can be influenced by handling the probe cable. Becoming a human radiator/capacitor by holding a finger to some parts of the board picks up a lot of unwanted noise. The PRP1 waveform is unchanged and robust.
-
Mine arrived the other day and I had a quick moment to get familiar with it today by doing some quick measurements.
Many thanks for the detailed write up, I'm glad you find it useful.
I use "Custom" attenuation on the oscilloscopes that I tested the unit on. If that's not supported, than a Math channel could be used for the probe attenuation. I think on some scopes you might even add the DC offset in with the Ax+B computation, and get the correct values on the screen.I noticed during the gain test that my sig-gen (DG1022) wasn't too great at hitting the Vpp setpoints. The output BNC was into a tee then connected to the PRP1 and CH2 as reference. I'd expect relative
Signal generators expect 50 Ohm load on their output. The PRP1 load is never really 50 Ohm, it's somewhere between 60-50KOhm depending on the frequency, by design (that's the whole point of it). I had a bit of trouble making the test setup first, until I realized I would need a source with 0 Ohm output, or do a relative measurement. -
I use "Custom" attenuation on the oscilloscopes that I tested the unit on. If that's not supported, than a Math channel could be used for the probe attenuation. I think on some scopes you might even add the DC offset in with the Ax+B computation, and get the correct values on the screen.
I don't believe I have a custom attenuation setting unfortunately, though I should try abusing probe compensation settings.
Ax+B isn't a great on this scope and most of my troubleshooting or tuning can tolerate relative measurements most of the time.Signal generators expect 50 Ohm load on their output. The PRP1 load is never really 50 Ohm, it's somewhere between 60-50KOhm depending on the frequency, by design (that's the whole point of it).
Yeah. I had the sig-gen set to 50R output to drive the scope's 50R, and basically just let the PRP1 hang off the tee. It should only be a few milliohms off so I didn't expect >5% error on the scope Vpp.
Saying that, 8-bit scopes aren't great tools for precision voltage measurements and the arb probably needs a closer look because it's spent a few years acting as a glorified TTL/LVCMOS pulse generator.
[EDIT]: Just checked the DG1022Z's 50R output spec and it's ±(1% of the setting value + 5mV + 0.5% of the amplitude) which explains it. Sorry for the off-topic clutter.