Test Equipment Anonymous (TEA) group therapy thread

[bd139]

Chief drunken monkey in my case.

[Specmaster]

Oh I see, well then in that case are you sure that the noisy floor is not just a case of blurry eyes from the night before 

[bd139]

quite possible

[Cerebus]

They are like slide rules: schoolkids and undergrads have 12" slide rules, grad students have 24" sliderules, engineers have up to 34' sliderules (note the change of units!), but professors and chief engineers have 6" sliderules.

And ground breaking engineers have the back of a fag-packet/beermat/a page out of the NY Times.

[bd139]

My big idea is on the back of a Wagamama table mat so fingers crossed that’s true

[Specmaster]

My big idea is on the back of a Wagamama table mat so fingers crossed that’s true

Ohno, thats the one you left behind on the table 

[URI]

...and now for something completely different! 

A new Test Equipment hit my bench: The Fluke PM6681 was delivered finally!
And for though I don't really need it and it wasn't advertised I'm lucky I got a unit with a high precision OCXO that has proved accuracy and stability by now. The instrument was also packed well in a double-layer cardboard package. 

BUT: 
On arrival the instrument had a thermal issue: After powering it up it was working nearly normal only for completely showing rubbish after warming up for a while.
Even the "check" function that counts the internal 100MHz source with that as its own reference showed unreasonable values jumping around. On top of that vibrations made them dance even wilder. 

What to do first? Contacting the seller to get (partial) refund? Or look inside and see if it's something serious?
I took the wrong decision. I got to my bench and cracked it open. 

First thing I found out was that it's really a thermal issue: I took a hefty 120x120mm² fan and cooled the bare PCB. After a short while the instrument started to work nearly normal -means: stable and plausible- again! Removing the fan restarted the game.
The self test threw an error for ASIC No 2...

A quick look over the PCB (upside and bottom side) showed some earlier rework. Not bad but not really good: solder residue and thick solder points. The rest of the board had a (original) solder work that didn't really look good as it was oxidised.
I found the first cold solder joint after a few minutes. But resoldering it didn't heal the instruments behaviour. Oh no..

I studied the service manual and identified the ASICs and their surrounding circuits and after poking around confirming I was on the right way I began to rework all solder joints of the TH parts -I had seen that some of the parts legs were cut after soldering -bad decision, everyone knows now -but back when it was manufactured..?
I took my magnifying glass and began to check the whole board. A lot of solder points looked like they could be cold ones. 
I reworked 2/3 of the questioned solder joints and the behaviour of the counter changed!
It was now stable unstable..   
It was after 3 o'clock in the morning when I decided to catch some three ours of sleep before getting up for work again. Boy, I was frustrated at that point!

After work I immediately got home and went straight to my bench and finished reworking the rest of the TH solder joints.
I was sceptical if that would bring the counter back to normal life but I had a rest of hope. 
So I plugged in mains power -fingers crossed- and observed the display..
It showed stable 100.000 000 0 MHz. I was relieved that it worked normal again and not stable unstable as before. But no thumbs up at that early point of time. I have waited for several ours now and the counter still works normal and stable. 

It works even more stable than my PM6680 and PM6685!
I think I'll have to examine whether its due to the higher resolution of the PM 6681 or due to upcoming cold solder joints in the PM6680/6685 now.. 

But for the PM6681: Winner! 
-in the first place. I gave away the chance of getting a refund and repair afterwards. 
I'm curious if the seller will ask me for feedback and if he will reveal some knowledge of the counters faultiness   

PM: I wrote "nearly normal" to describe the behaviour of the PM6681 counter: After reworking the TH solder joints of the analog part the counter worked even more stable and accurate than it did before in the first minutes warming up.

I'm away now to make up for the missing hours of sleep..

[bd139]

Nice job. You made the right decision 

I really hate that feeling when you crack something open and find someone else’s rework that was done badly. It’s just the worst thing in the world.

[SparkyFX]

Claim has already been submitted, can't make a decade box with that combination.

Build a capacitance decade then? Works with BCD.

[Specmaster]

Hmm, I think that I would have contacted the seller first to see what they had to say about. You might have got a complete refund, especially if the claimed that it was working in the description of the unit.

The you would have had the satisfaction of doing the repair and that lovely sweet feeling of having gotten yourself a real bargain and a half. Anyway well done to you, I suppose that you could still have that feeling if the price was right in the first instance was it? 

[Specmaster]

Claim has already been submitted, can't make a decade box with that combination.

Build a capacitance decade then? Works with BCD.

Well I might just do that later, if I get a refund, trouble is though that I took delivery of the 1% and .1% resistors and a suitable enclosure from Mouser yesterday in readiness for the switch banks arrival. I did also however order another thumbwheel switch bank from another veller as I did have my doubts about the 1st seller. They showed pictures of a BCD switch but stated it was decimal  in the description. The other seller did not show a good clear picture of the terminals but the description claimed it was a decimal unit. I ordered them both on the same day so there's a good chance it could arrive tomorrow, will it be the same... 

I'm beginning to wonder now if they they thought that because it shows 0-9 on the face when you click it round, that they thought it was a decimal and that a BCD one would show BCD??   

[bd139]

Just a gotcha with those switches you need to be aware of. The contact resistance can be around the 0.5-1 ohm mark and varies with pressure and quality. The ones around the £14 a switch mark have a guaranteed 0.1 ohm resistance under all conditions. If you cascade several together, say 8, then you introduce 4-8 ohms error which makes a 1 or 10 ohm range a little problematic from a readout perspective. I wouldn't bother going below 10 ohms or expecting terrible levels of accuracy. Fine for 100 ohms and above as error isn't significant. They are fundamentally designed for digital levels where the precise impedance isn't a problem.

Been playing with automation this evening to exercise my SCPI+python skills. Got the U1241C and GDM-8341 hooked up to a diode and rather boring series resistor and using DG1022Z and a sneaky hack I found (set it to 1uHz pulse, never trigger it and change the offset and you've got a programmable +/-5V power supply!). Sweep the offset, record the V and I and chuck it in matplotlib.

Dead simple but good fun. One digital curve tracer!

Output:



Ordered some light bulbs to play with tomorrow.

Edit: there is a purpose for this: matching diodes for diode ring mixers.

[neo]

Happy thanksgiving, TEA.

[Specmaster]

Yeh, I was a aware that the contacts arent all that but it seems that people who have already build decade boxes with them get a typical 1 to 1.5 ohms resistance built in, which is more then livable with, so if for example you wanted 500 ohms then you just dial up 499 and you can always trim and check it with a DMM to get it really close to the desired value. But as you rightly say, in the units is where the accuracy is really going to suffer. That being said though, its main purpose currently is a means of checking the calibration of the DMM's along with the voltage reference as it is more likely that the readings I'll taking with them is going to resistance and DCV above all else.

Glad to see that you've discovered another use for that Rigol in the short time its got left with you before departing for pastures anew.

Did you email Keysight at all about dropping to "poke" their scopes?

[bd139]

That's not terrible then. I could live with that.

I haven't emailed them yet. I'm waiting until Feb/Mar. Ideally I'd like to fund it via buy/sell on ebay as I get to keep the moral high ground on spending then

[Specmaster]

That makes sense, your bookkeeping skills will be put to good use there and you might just miss out on the old compo gift buying ploy there, especially if you point out her "pin" money opportunities with her hairdressing skills 

[Specmaster]

Here's a link to the switches I ordered and they are clearly marked up as decimal https://www.ebay.co.uk/itm/10x-KSA-2-Decimal-10-Position-Single-Unit-Thumbwheel-Switches/121745184682?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649

Then there are these ones that are clearly marked as BCD encoded https://www.ebay.co.uk/itm/10-x-Black-BCD-Code-Single-Unit-Thumbwheel-Pushwheel-Switches-16mm-x-6mm-KM2/291583344055?_trkparms=aid%3D111001%26algo%3DREC.SEED%26ao%3D1%26asc%3D20160908105057%26meid%3D1b31268592eb484e85d863626697a805%26pid%3D100675%26rk%3D2%26rkt%3D15%26sd%3D172904665517&_trksid=p2481888.c100675.m4236&_trkparms=pageci%253Ae6b41782-d0a6-11e7-a74c-74dbd180916a%257Cparentrq%253Aeb3be20815f0a9c10cb72fa9fffa5ea1%257Ciid%253A1

These are from the same seller so they must have made a mistake and I've requested that they send me out urgently the correct item, especially as I paid a premium price for them.

So glad I did this check because I was beginning to wonder if I had made a mistake when ordering them, which I didn't. 

[montemcguire]

Case in point, my DG1022Z arb has an interesting problem where at very slow (200mHz) square waves it gives sinusoidal wiggles on the leading edge for no apparent reason. Makes it bloody useless as a pulse generator. Waiting for a 33120A to come along instead.

Ripply pre and post ringing is most likely "Gibbs phenomenon", which is caused by an arbitrary waveform synthesizer trying to produce a square wave with a necessarily bandlimited DAC. If the ripples are worse at very low frequencies, they probably reduced the DAC's sample rate to cope with limited waveform memory, a problem that could be solved by throwing money at it, but which won't happen so often with budget gear.

Long and short, arb generators are great for most everything except for pulses or square waves. Use a real 'native' pulse generator for pulses and get happier! :-)

[URI]

Hmm, I think that I would have contacted the seller first to see what they had to say about. You might have got a complete refund, especially if the claimed that it was working in the description of the unit.

Yeah, I should have done it. 

The you would have had the satisfaction of doing the repair and that lovely sweet feeling of having gotten yourself a real bargain and a half. Anyway well done to you, I suppose that you could still have that feeling if the price was right in the first instance was it? 

The price (330€) was good for an instrument of this class -300MHz universal counter with 10digits display and 12 digits resolution using Math functions (subtracting an offset to show 2 extra digits) without issues. Ok, I gave away that chance for the immediate adventure of exploring and repairing. 
It's still not so bad at all but by now I regret passing up the opportunity of getting a refund to trade off the multiple cold solder joints that had to be reworked to make the counter work stable again.

And I think it still has minor issues: removing the reference frequency from the according input lets the counter still work from that input obviously taking up noise of the internal reference and perhaps the 100MHz source -resulting in a huge error of around 240 kHz at 10MHz input signal. As the circuit of switching between the external and the internal reference is done by simple 74.. Logic and some transistors this should be a problem not too hard to solve I hope.

[HalFET]

Case in point, my DG1022Z arb has an interesting problem where at very slow (200mHz) square waves it gives sinusoidal wiggles on the leading edge for no apparent reason. Makes it bloody useless as a pulse generator. Waiting for a 33120A to come along instead.

Ripply pre and post ringing is most likely "Gibbs phenomenon", which is caused by an arbitrary waveform synthesizer trying to produce a square wave with a necessarily bandlimited DAC. If the ripples are worse at very low frequencies, they probably reduced the DAC's sample rate to cope with limited waveform memory, a problem that could be solved by throwing money at it, but which won't happen so often with budget gear.

Long and short, arb generators are great for most everything except for pulses or square waves. Use a real 'native' pulse generator for pulses and get happier! :-)

Was thinking you could actually use a cheap simple cpld or fpga between the memory and DAC to remedy this quite easily, but even that was probably out of budget for rigol.

[Berni]

Was thinking you could actually use a cheap simple cpld or fpga between the memory and DAC to remedy this quite easily, but even that was probably out of budget for rigol.

Siglent some time ago released series of signal generators that claim to have solved the problems:
http://www.siglent.com/ENs/pdxx.aspx?id=1134&T=2&tid=16

I have been thinking of buying one but they do still cost a pretty penny for a Chinese piece of test gear. But then i managed to score a old Tek AWG2041 for the right price. The things was the bees knees back in the day and doesn't suffer from the usual DDS issues due to having a widely adjustable synthesized frequency for the DAC clock. Tho 1GS/s DAC technology had a bit of a way to go back then so its only 8bit

[bd139]

Interesting comments on the DG1054Z.

I’ve replaced the pulse generator role with a couple of 74HC14’s, three caps and a couple of trimmers to set width and rate. Not very precise but good enough!

[djos]

Chaps, how useful is an old single channel 5 megahertz GW oscilloscope these days? It's not even dual trace but for $50 I'm kinda tempted.

[bd139]

I wouldn’t pay $50 for it. $1/MHz is a sensible price.

I think the last scope I had of that bandwidth I paid £5 (about $6 or $7) and that had a box, probes, manuals and was dual trace.

[djos]

I wouldn’t pay $50 for it. $1/MHz is a sensible price.

I think the last scope I had of that bandwidth I paid £5 (about $6 or $7) and that had a box, probes, manuals and was dual trace.

Sounds like a good rule of thumb, we get stiffed pretty badly here in Aus tho so everything is over priced.

Thank dog Amazon is finally launching properly here next week!