Test Equipment Anonymous (TEA) group therapy thread

[Robert763]

Yeah saw that one. At that price, nope!

@Specmaster: nice. I’ve never seen inside one of them. Seems to be standard bits which is good 

Just doing full check out on 5381a and some RTFM and this is actually a really nice counter. Despite having only 7 digits it has a nice overflow system so you can use it down to 0.1Hz resolution at 80MHz.   It it appears the decimal points on KHz and Hz are not working though. So some debugging required. These are dumb switched ones so I reckon it’s the front panel switch or LEDs which have failed. Someone has already replaced three of the displays so may be related. At least I’ve got something to do now!

Well it didn’t take a long time to work out the problem with the counter decimal point. The idiot who replaced the right three displays put the wrong ones in  . Have attached the utterly annoying disparity.

I’ve located the data sheet for the original part (5082-7731) and located a suitable replacement (Kingbright SA03-11RWA). Now thinking if I should do all 7 or just the wrong three.

I may have a set of LEDs do you want me to look?

[mnementh]

Looking at the pricing, it appears they're pretty much globally cheaper by 10 than by 7. Lots of spares then.

mnem
 

[Cubdriver]

Nor me.  Good luck!!

-Pat

<edit to add - didn't notice the thread had rolled to the next page - this is in regards to the transfer oscillator discussion between the fregg and David>

[ch_scr]

Finally, I found the source of much grief: The reference buffer that supplys the two (hopefully identical) signals to the measuring device. Somehow I decided it would be a good idea to go with "combine 2 inverters per output channel", whilst beeing all tied together on the inputs. As it turns out, that induced a lot of random jitter between the channels. Enough to make looking at noise performance of my actual device under test a boondoggle. The new topology splices all inverters together through individual series resistors and the outputs are hooked up like a resistive splitter. This way, the signal has a nice impedance and to be very similar on both channels (given both are terminated into 50 ). At least thats what I hope.

To not make more of a fool of myself I continued to actually measure the new distribution, sadly no pictures from before.

As can be seen they are close. The -86ps should be delay from the length difference in the sma cables used for hookup.
TIL: Logic gates can introduce random jitter. A lot more than I thought 
Although I am not sure, but cycle-to-cycle jitter still looks not so good for a precision 10MHz source? Any opinion on that?
To be specific I fear my "unbuffered inverter as wideband limiting amplifier" implementation might have bad performance in that regard?
Edit: Oh FFS one of the outputs has fallen off 

[bd139]

I may have a set of LEDs do you want me to look?

Thanks for the offer but have already ordered some from Rapid 

[bd139]

Finally, I found the source of much grief: The reference buffer that supplys the two (hopefully identical) signals to the measuring device. Somehow I decided it would be a good idea to go with "combine 2 inverters per output channel", whilst beeing all tied together on the inputs. As it turns out, that induced a lot of random jitter between the channels. Enough to make looking at noise performance of my actual device under test a boondoggle. The new topology splices all inverters together through individual series resistors and the outputs are hooked up like a resistive splitter. This way, the signal has a nice impedance and to be very similar on both channels (given both are terminated into 50 ). At least thats what I hope.

To not make more of a fool of myself I continued to actually measure the new distribution, sadly no pictures from before.

As can be seen they are close. The -86ps should be delay from the length difference in the sma cables used for hookup.
TIL: Logic gates can introduce random jitter. A lot more than I thought 
Although I am not sure, but cycle-to-cycle jitter still looks not so good for a precision 10MHz source? Any opinion on that?
To be specific I fear my "unbuffered inverter as wideband limiting amplifier" implementation might have bad performance in that regard?

This is interesting as I was literally just about to do this. I get the feeling my demands are somewhat less than yours however so may not be an issue. I need a 1MHz and 10MHz source so was going to use my emergency OCXO (everyone needs an emergency OCXO!), a 74HC390 to /5 then /2 and buffer both the 10MHz and 1MHz output with a 74AC14.

[mnementh]

Today's Project: Cheap RF Active Probe      https://a360.co/37CE8El   

...Next on the agenda will be to do some testing, once I find appropriate adapters. Maybe I'll faff around with it on my TinySA/nanoVNA as well. 

And maybe I should add a probe cover; that little fucker is poky! 

Woke up at 6AM; this was stomping around the back of my head like Tim Curry in knee-high boots  and wouldn't let me go back to sleep.     Now printing; due to hollow spaces, it actually needed to be printed vertical orientation. Should be interesting; I probably screwed up at least 3 things on it, as I didn't even make my coffee until the file was done and printer warming up. 

   

Well, fuck-a-doodley-doo... Whadda I do widd m'self NOW...?!?

Damned thing... it all fits perfectly.    Even the battery holder is a perfect slightly snug fit. 

mnem
*toddles off to drag out that Pioneer receiver*

[Robert763]

I may have a set of LEDs do you want me to look?

Thanks for the offer but have already ordered some from Rapid 

Just as well I checked and the ones I have are 5082-7731 (RHDP) so no good.

[mnementh]

...I need a 1MHz and 10MHz source so was going to use my emergency OCXO (everyone needs an emergency OCXO!), a 74HC390 to /5 then /2 and buffer both the 10MHz and 1MHz output with a 74AC14.

Only in the TEAnonymous thread... 

mnem
...and certain rings of hell.

[bd139]

So much nope!

https://www.eevblog.com/forum/buysellwanted/agilent-34401a-286199/

[ch_scr]

This is interesting as I was literally just about to do this. I get the feeling my demands are somewhat less than yours however so may not be an issue. I need a 1MHz and 10MHz source so was going to use my emergency OCXO (everyone needs an emergency OCXO!), a 74HC390 to /5 then /2 and buffer both the 10MHz and 1MHz output with a 74AC14.

Yeah this thing sure does amplify fine time differences 
I have one output to the side from a different gate so I can measure a comparison to that case - but not before sunday evening.
Will report back, now I'm giddy to see if I can measure a difference with the scope there.

[bd139]

It is certainly feasible that gate propagation times are different enough for this to be an issue. I look forward to your results

[Wolfgang]

...I need a 1MHz and 10MHz source so was going to use my emergency OCXO (everyone needs an emergency OCXO!), a 74HC390 to /5 then /2 and buffer both the 10MHz and 1MHz output with a 74AC14.

Only in the TEAnonymous thread... 

mnem
...and certain rings of hell.

It works, but its a bodge. A solution with a little less jitter is to use a *synchronous* divider IC.
The really posh people do it via an injection locked subharmonic divider, of course.
Anyway, this is a thread for addicted people so everything goes. 

[ch_scr]

I want to try to better describe what the device intends: When comparing two voltage references, you might want to make a differential measurement between them, because your voltmeter shows the same measurement with regard to measuring both voltages to ground. For an oscilloscope you would want to use a differential amplifier in such occasions. Now imagine this, but for frequency references and a frequency counter instead of the voltmeter and voltage references. 

[bd139]

I want to try to better describe what the device intends: When comparing two voltage references, you might want to make a differential measurement between them, because your voltmeter shows the same measurement with regard to measuring both voltages to ground. For an oscilloscope you would want to use a differential amplifier in such occasions. Now imagine this, but for frequency references and a frequency counter instead of the voltmeter and voltage references. 

That's where your expensive tens of GHz oscilloscopes win. They have the ability to discriminate picosecond phase shifts. Assuming all your probes are exactly the same length and a thousand other small parameters . Also some frustrating things to consider: propagation of signals through silicon are a mere fraction of c and are not necessarily consistent depending on the deposition process used.

Ergo best to squint a bit and pretend the world is blurrier than it actually is and be satisfied with it 

Edit: more chaos as well. You're probably worrying about 10MHz or so. 1000-3000GHz band is open here if you file a NoV with Ofcom 

[DC1MC]

Depressed that my car brakes need replacement I've bought 500 of these HLMP-ED31-QT000 for 8euroni. If someone needs Agilent original LEDs, let me know 

[bd139]

Light them all up and make your place look like a brothel 

[DC1MC]

Light them all up and make your place look like a brothel 

I need one of these PCB with a gadzillion of LEDs and some drivers to display pictures, or you mean the brakes ?

[mnementh]

Light them all up and make your place look like a brothel 

I need one of these PCB with a gadzillion of LEDs and some drivers to display pictures, or you mean the brakes ?

I can do both.

mnem
...at the same time, if you give me a few minutes to prepare.

[mnementh]

...Ergo best to squint a bit and pretend the world is blurrier than it actually is and be satisfied with it 

Good advice for a multitude of scenarios. 

mnem
Tito's is my preferred blurring agent. and flux remover.

[bd139]

Gaaaaaah fuck the universe. Just spent 2 damn hours trying to get DisplayPort audio working on my PC. Then I trip over this nugget: https://www.asus.com/support/FAQ/1045437/ .... Just magic reset your BIOS. It works?!?!?!?! But why?!?!?!?!!

I am seriously not impressed with this Asus board. Hardware is fine but the firmware is awful. ASRock was better.

[dl6lr]

Gaaaaaah fuck the universe. Just spent 2 damn hours trying to get DisplayPort audio working on my PC. Then I trip over this nugget: https://www.asus.com/support/FAQ/1045437/ .... Just magic reset your BIOS. It works?!?!?!?! But why?!?!?!?!!

I am seriously not impressed with this Asus board. Hardware is fine but the firmware is awful. ASRock was better.

Asus seems to be gamer crap now. They used to have good boards back then, but they fucked up. My son got lost with incompatibilities between an Asus board and Asus graphics. He dumped the board and bought one from Intel.
If you want to have firmware trouble, buy Dell 
My employer orders Dell hardware. Now every time I boot my Laptop, it gets stuck with 0.8GHz maximum speed. To cure the issue, you have to unplug the power from the docking station and replug it after 15 seconds.    Maybe they'll fix it with a firmware update to break other things.

[mnementh]

You can't really blame the board/BIOS just because someone or some installer changed a setting in the BIOS. This is the weakness of UEFI; you give control of the BIOS to the OS, it's gonna throw all kinds of shit in there.

mnem
 

[Wolfgang]

So more tips to get this clean:

- decent PCB with a groundplane
- individual PSU paths to the ICs with separate filtering (RC). Reduces crosstalk via power lines
- Low noise regulator (LM723 instead of LM7805). PSU noise is directly translated to switching jitter.
- No double functions in a package. There is always crosstalk via supply lines within the chip.
- Longer PCB tracks: use terminations
- No faster logic than you need. And no Smith triggers, please. Just measure supply current spikes and you know why.
  So output over HC04 is good enough. And one package per output frequency.

Some info can be learned from how NIST make their dividers.
 

[Kosmic]

By the way, congrats for your PhD Wolfgang. Quite an accomplishment.