Test Equipment Anonymous (TEA) group therapy thread

[factory]

Am I missing something here? someone thinks a 1960's HP 6269A (40V 50A PSU) is worth nearly £500 (plus almost £1/kg for the courier). 

eBay auction: #184164186009

P.S. Sticking a later HP badge on the front will not fool me.

[med6753]

The 535A has been powered up for about 2 hours and everything looks OK.  The slight increase in B+ voltage has futz'ed with the DC balance in the CA plug-in. That's not surprising, it's very sensitive. I'll reign it back in after I get all the caps done and I'll also check vertical gain and a few other cal points.

I plan on ripping into it again tomorrow.

[mnementh]

   Just stumbled across this while looking at arcade joysticks on fleaBay: eBay auction: #132490173176   Not sure if it's valuable as HAM gear or not; just thought I'd bring it to the attention of those who are into aggravated aether.

mnem
*back to the bench*

most jurisdictions rather frown on unlicensed 3M operation

Key Parametrics ：
Frequency (Min) (MHz)              1.8
Frequency (Max) (MHz)            150
Supply Voltage (Typ) (V)            50
P1dB (Typ) (dBm)                      60
P1dB (Typ) (W)                       1000
Output Power (Typ) (W) @ Intermodulation Level at Test Signal  1000.0 @ Peak

UP TO 150MHz; testing optimized at 130MHz/2.3m. Min frequency range goes down well into the 160m band, but no idea of performance at those frequencies. 

Nothing about input or output impedance and also: https://www.jhhhhjonline.com/index.php?main_page=product_info&products_id=372552

Since when has that mattered to a true HAM...? Matching impedances is literally half the game. 

Description specifies that the module is unmatched, meaning you have to calculate/design that yourself depending on application. A quick Google brings up this white paper from the folks who manufactured the final output semi: https://www.nxp.com/docs/en/application-note/AN1526.pdf Interesting, but just skimming it I can see it's way over my head. 

However, since this module is well... a module, intended to be dropped into industrial/commercial devices jellybean-style; I'd expect it should be designed to interface pretty simply with whatever matching network you design.

Great deal finding it cheaper elsewhere.  As I said; not sure if it has HAM value or not... it just seemed like a lot of aggravated aether in a little package. 

mnem
*currently stuffing boogers in a baggie*

[WastelandTek]

that pallet is 50 ohm for sure   broadband coaxial transformer match

[Kosmic]

The 1N829 Zener diode is temp compensated and do not need an oven. TC has been measured when the unit was produced and is compensated on the board (can be changed by soldering / unsoldering some solder bridges). The Solartron 7081 (8.5 digits) also has the ref placed on the board like that with no protection. But in the case of the 7081, there's a temperature sensor on the board used to correct the voltage reference dynamically.

Nevertheless, it’s true that the transformer is getting really hot. I always wonder if it was always like that or this is something that appeared over time ?

In term of noise, on my 7061, I had a lot more success when I readjusted the internal clock. The procedure is described briefly at the end of the maintenance manual.   

I get 2.91V -  2.97V at TP402, so I should be good there.    I know that the Zener current is hand adjusted to match each specific diode to set it to the 0 tempco point, and that's all well and good. What that adjustment can't help is when a temperature differential is created over the diode body or the diode's leads.

The transformer really does get quite warm. With the lid off, I'm measuring over 42°C. Then again, my 7150 gets just as hot, so it's probably quite normal.     I think I'll see if I can get my thermocouple probe in through the earth screw on the back and see what the temperature looks like around the reference area once the meter is closed again and warmed up.

Have you seen this suspicious-looking solder joint of the 1N829A?

It does look a little suspicious - you have eagle eyes! Looking closer, the leads appear not to have been soldered on the top of the board. My assumption is that they hand soldered the reference after the rest of the PCB had been flow soldered and probably used a heat sink on the top of the PCB while they soldered the legs on the bottom side.

  Certainly, given the current reasonable accuracy that the meter exhibits, I don't have any reason to believe that there is any problem with the connections to the reference diode.

It seems to me that while the clock reference procedure is important, grizewald has presented valid empirical evidence that applying insulation around the reference diode improves overall stability.

When noise, and thermal emf is involve you need more than a spot check. Logging the same ref (or simply a short) for 2 days with and without isolation would probably be interesting.

[mnementh]

   Submitted for your TEApproval: A Baggie of Boogers   

                                     

I think I may have reached a new low... I just spent 10 minutes stuffing glue-boogers rubbed off a piece of panel into a baggie to save for later; just because they're so good at cleaning bits of magnetic dirt from magnetic things like these charging cables. 

mnem

[Kosmic]

It seems to me that while the clock reference procedure is important, grizewald has presented valid empirical evidence that applying insulation around the reference diode improves overall stability.

Thermal insulation to reduce thermal drift. Lock the clock to the mains frequency to reduce power line noise.

As the "thermal" in question is generated by the device itself, not exactly thermal drift; hence my classifying it as a stability issue. I was trying to point out that the two issues are not an either/or solution as Kosmic's post sortof sounded... however in retrospect, I know Kosmic is generally not that kind of nit-picky, so I agree here in principle.

My understanding of the theory is that isolating the voltage ref was going to reduce noise associated to thermal variation inside the case. And that it look like an oversight from the designers (Solartron). That's why I was pointing to the clock which is the best way of reducing noise.

Now it's an interesting theory and it probably worth testing over multiple days.

In any case, I think doing maintenance on old equipment to reach original specs is one thing. Improving the original design is something else. I'm pretty sure the engineers at Solartron were a lot more competent than me on that topic.

[med6753]

I think I may have reached a new low...
mnem

Yes, you have. But I think it's just wonderful. 

[mnementh]

...In any case, I think doing maintenance on old equipment to reach original specs is one thing. Improving the original design is something else. I'm pretty sure the engineers at Solartron were a lot more competent than me on that topic.

Yeah... this is why I prefer to enjoy OPVN vicariously. I have enough nutteries of my own; I already can't keep up. 

mnem
No, I may not touch it.

[worsthorse]

   Submitted for your TEApproval: A Baggie of Boogers   



I think I may have reached a new low... I just spent 10 minutes stuffing glue-boogers rubbed off a piece of panel into a baggie to save for later; just because they're so good at cleaning bits of magnetic dirt from magnetic things like these charging cables. 

mnem

don't show that to anyone as it is likely to provoke an intervention. 

[mnementh]

 

Especially for any parent who's ever had a sniffly toddler...

mnem
"Oh gawd... doesn't this river of slime ever end...?"

[Mortymore]

Just realized, many of TEA regulars here, own Tek 24x5 A/B/plain, aren't we ?

I'm a mere observer here, but I have my admission pass 

[bd139]

I’ve actually got some 74LVC14’s floating around but my eyes aren’t up to that this evening

I may build something decent based on your suggestion next week. Need a non radio project!

You need very good decoupling. Work out the di/dt into 50ohms, then the induced voltage due to any stray inductance v=Ldi/dt. The numbers aren't small!

Yeah that will explain the crappy results I have got here. This was more a quick test while I hum and haw around buying a Bodnar pulser.

So quick 74AC14 pulser. Poor decoupling and lots of inductance so we're not expecting miracles here. 2ns is the order of the day which is ~175MHz. Need faster really for this crate.



Yes that's a horrible pulse isn't it (yes the 100% mark is actually the out high voltage ... eventually)



And a really terrible attempt at TDR on the end of about 5 foot of fail:



Added to shopping list: http://www.leobodnar.com/shop/index.php?main_page=product_info&products_id=295

This little device will be relegated to other tasks. It's still useful as it stands.

[Wolfgang]

I’ve actually got some 74LVC14’s floating around but my eyes aren’t up to that this evening

I may build something decent based on your suggestion next week. Need a non radio project!

You need very good decoupling. Work out the di/dt into 50ohms, then the induced voltage due to any stray inductance v=Ldi/dt. The numbers aren't small!

Yeah that will explain the crappy results I have got here. This was more a quick test while I hum and haw around buying a Bodnar pulser.

So quick 74AC14 pulser. Poor decoupling and lots of inductance so we're not expecting miracles here. 2ns is the order of the day which is ~175MHz. Need faster really for this crate.



Yes that's a horrible pulse isn't it (yes the 100% mark is actually the out high voltage ... eventually)



And a really terrible attempt at TDR on the end of about 5 foot of fail:



Added to shopping list: http://www.leobodnar.com/shop/index.php?main_page=product_info&products_id=295

This little device will be relegated to other tasks. It's still useful as it stands.

The art brut of nanosecond pulse hacking. You need not be ashamed - the Jim Williams stuff looks almost as fancy.
I mean - with some ECL or 10G gate stuff everybody can do it.

[Cubdriver]

Just realized, many of TEA regulars here, own Tek 24x5 A/B/plain, aren't we ?

Guilty as charged...



-Pat

[WastelandTek]

quick test while I hum and haw around buying a Bodnar pulser.

just get Leo's pulser, makes everything else obsolete

[grizewald]

When noise, and thermal emf is involve you need more than a spot check. Logging the same ref (or simply a short) for 2 days with and without isolation would probably be interesting.

Yeah, I'm about half way through battling with a GPIB cable soldering pin headers onto various wires so that I can plug it into an Arduino and get some logging going. Also just finished getting a collection of Wemos boards running temperature, humidity and pressure sensors and sending the measurement once every <sample period> to an MQTT broker running on my NAS.
Then I'll need some shell script to combine the readings from the environmental sensors with the voltage readings from the 7061 along with a timestamp for each reading.

The only problem is that it's nearly midnight and there's this inconvenient thing called work tomorrow.

But I will be recording some long term statistics from the 7061 as soon as I have that all done. Now if only I could get a GPIB adapter for my 7075...

[beanflying]

It seems to me that while the clock reference procedure is important, grizewald has presented valid empirical evidence that applying insulation around the reference diode improves overall stability.

Thermal insulation to reduce thermal drift. Lock the clock to the mains frequency to reduce power line noise.

As the "thermal" in question is generated by the device itself, not exactly thermal drift; hence my classifying it as a stability issue. I was trying to point out that the two issues are not an either/or solution as Kosmic's post sortof sounded... however in retrospect, I know Kosmic is generally not that kind of nit-picky, so I agree here in principle.

My understanding of the theory is that isolating the voltage ref was going to reduce noise associated to thermal variation inside the case. And that it look like an oversight from the designers (Solartron). That's why I was pointing to the clock which is the best way of reducing noise.

Now it's an interesting theory and it probably worth testing over multiple days.

In any case, I think doing maintenance on old equipment to reach original specs is one thing. Improving the original design is something else. I'm pretty sure the engineers at Solartron were a lot more competent than me on that topic.

Sometime ago I asked a similar question about the LM399 Vrefs in my 34970A's which only got a might be worth it response from a few of the Metrology heads https://www.eevblog.com/forum/metrology/hp-agilent-34970a-precision-logging-thread/msg1377876/#msg1377876

Keeping direct air movement away would be more important than insulation IMO so consider a dome over it without the insulation and let the temperature sit where it was designed. For those without a 3D printer https://www.amazon.co.uk/Assorted-Fillable-Plastic-Surprise-Easter/dp/B01N9ZM9SK for those who sensibly have one you know what to do 

[Kosmic]

It seems to me that while the clock reference procedure is important, grizewald has presented valid empirical evidence that applying insulation around the reference diode improves overall stability.

Thermal insulation to reduce thermal drift. Lock the clock to the mains frequency to reduce power line noise.

As the "thermal" in question is generated by the device itself, not exactly thermal drift; hence my classifying it as a stability issue. I was trying to point out that the two issues are not an either/or solution as Kosmic's post sortof sounded... however in retrospect, I know Kosmic is generally not that kind of nit-picky, so I agree here in principle.

My understanding of the theory is that isolating the voltage ref was going to reduce noise associated to thermal variation inside the case. And that it look like an oversight from the designers (Solartron). That's why I was pointing to the clock which is the best way of reducing noise.

Now it's an interesting theory and it probably worth testing over multiple days.

In any case, I think doing maintenance on old equipment to reach original specs is one thing. Improving the original design is something else. I'm pretty sure the engineers at Solartron were a lot more competent than me on that topic.

Sometime ago I asked a similar question about the LM399 Vrefs in my 34970A's which only got a might be worth it response from a few of the Metrology heads https://www.eevblog.com/forum/metrology/hp-agilent-34970a-precision-logging-thread/msg1377876/#msg1377876

Keeping direct air movement away would be more important than insulation IMO so consider a dome over it without the insulation and let the temperature sit where it was designed. For those without a 3D printer https://www.amazon.co.uk/Assorted-Fillable-Plastic-Surprise-Easter/dp/B01N9ZM9SK for those who sensibly have one you know what to do 

LM399 is different since it has it's own oven and air draft protector built in (you can actually buy the chip without the white plastic cover). The only instrument I seen encapsulating the LM399 is the Valhalla 2720GS where 2 LM399 are inside an oven. They are probably running the 1st oven a bit cooler to compensate.

[Kosmic]

When noise, and thermal emf is involve you need more than a spot check. Logging the same ref (or simply a short) for 2 days with and without isolation would probably be interesting.

Yeah, I'm about half way through battling with a GPIB cable soldering pin headers onto various wires so that I can plug it into an Arduino and get some logging going. Also just finished getting a collection of Wemos boards running temperature, humidity and pressure sensors and sending the measurement once every <sample period> to an MQTT broker running on my NAS.
Then I'll need some shell script to combine the readings from the environmental sensors with the voltage readings from the 7061 along with a timestamp for each reading.

The only problem is that it's nearly midnight and there's this inconvenient thing called work tomorrow.

But I will be recording some long term statistics from the 7061 as soon as I have that all done. Now if only I could get a GPIB adapter for my 7075...

If you have a VISA compatible gpib adapter I can send you my software. You configure via .xml file and it dump the data in .csv. I'm building the UI right now. Will eventually support graph, filtering and etc ..

[grizewald]

My understanding of the theory is that isolating the voltage ref was going to reduce noise associated to thermal variation inside the case. And that it look like an oversight from the designers (Solartron). That's why I was pointing to the clock which is the best way of reducing noise.

Now it's an interesting theory and it probably worth testing over multiple days.

In any case, I think doing maintenance on old equipment to reach original specs is one thing. Improving the original design is something else. I'm pretty sure the engineers at Solartron were a lot more competent than me on that topic.

I have to say that I'm in no way suggesting that I know better than the highly competent engineers at Solartron, far from it! Having seen how stable my 7075 is when measuring the virtually noise free Weston cell, I hoped the 7061 would be just as good but it seems to have much more noise.

You're absolutely right that some proper data logging with all the environmental data included would show exactly how good or bad the 7061 is. There's no substitute for some proper science!

I'm sure you'll agree though, from looking at the thermal image of the warmed up meter, that it's unlikely there is a stable thermal environment inside the case. Convection currents will roll around the inside constantly due to the uneven distribution of heat from the transformer. The fact that the reference is mounted high off the board on relatively long leads would suggest that local air currents could cause thermocouple effects.

I don't see any noise specifications in the Solartron manuals, so I have nothing to judge what is acceptable for the 7061. I know that the 7075 is about twice as accurate when comparing specs, even though it only has 14 million counts compared to the 7061's 21 million counts. Even so, I was surprised when I saw how it behaved with the Weston cell.

It may be that my suspicions are unfounded, but having the reference mounted high up off the board in a cool corner of a case with uneven internal temperature looks like a noise source to me. As you know, the 70[6,7,8]1 series meters all have an auto zero function which runs every 15 minutes to "correct for thermal drift" as the manual says. I think it's interesting that the older 7075 with its ovenised reference diode does not have this function and manages to deliver incredibly stable measurements to the microvolt level without it. Maybe there's some connection?

One other tool I'll be building to help look at the noise levels on all my precision meters is a xdevs.com KX LTZ1000A reference. I have all the parts now thanks to Mouser's incredible service. All I'm waiting for is the PCBs from OSHPark and some bismuth/tin super low melting point solder paste from China.

Rest assured that I won't let this remain as an unsupported assertion. Once my logging system is working I'll get the meter connected to my Weston cell for some initial testing with and without a tea cosy for the diode. I fear that the temperature sensitivity of the Weston cell may make the results hard to interpret, but if they do, the KX reference should make things clearer.

[tggzzz]

Now if only I could get a GPIB adapter for my 7075...

https://www.equipnet.com/auctions/catalog/uk-rf-testing-sealed-bid/1096/?AuctionSearch=7075

Particularly the first lot is as close as you will get

[grizewald]

If you have a VISA compatible gpib adapter I can send you my software. You configure via .xml file and it dump the data in .csv. I'm building the UI right now. Will eventually support graph, filtering and etc ..

I don't have such an adapter unfortunately, for two reasons: they cost way too much money for what they are; they use drivers which will only run on Windows and I won't use Windows on my machines at home.

I do have a Solartron BCD interface for the 7075, but I've never heard of anyone else who has implemented an Arduino interface for it. I can develop one myself with a Mega 2650 board as it has just enough I/O pins for the job (makes GPIB look like a serial interface when you get to counting wires), but it's not an insignificant project.

I'm kind of liking the idea of all my logging interfaces publishing their data to an MQTT broker. It can then be retrieved as needed in a platform neutral way for further processing.

[beanflying]

It seems to me that while the clock reference procedure is important, grizewald has presented valid empirical evidence that applying insulation around the reference diode improves overall stability.

Thermal insulation to reduce thermal drift. Lock the clock to the mains frequency to reduce power line noise.

As the "thermal" in question is generated by the device itself, not exactly thermal drift; hence my classifying it as a stability issue. I was trying to point out that the two issues are not an either/or solution as Kosmic's post sortof sounded... however in retrospect, I know Kosmic is generally not that kind of nit-picky, so I agree here in principle.

My understanding of the theory is that isolating the voltage ref was going to reduce noise associated to thermal variation inside the case. And that it look like an oversight from the designers (Solartron). That's why I was pointing to the clock which is the best way of reducing noise.

Now it's an interesting theory and it probably worth testing over multiple days.

In any case, I think doing maintenance on old equipment to reach original specs is one thing. Improving the original design is something else. I'm pretty sure the engineers at Solartron were a lot more competent than me on that topic.

Sometime ago I asked a similar question about the LM399 Vrefs in my 34970A's which only got a might be worth it response from a few of the Metrology heads https://www.eevblog.com/forum/metrology/hp-agilent-34970a-precision-logging-thread/msg1377876/#msg1377876

Keeping direct air movement away would be more important than insulation IMO so consider a dome over it without the insulation and let the temperature sit where it was designed. For those without a 3D printer https://www.amazon.co.uk/Assorted-Fillable-Plastic-Surprise-Easter/dp/B01N9ZM9SK for those who sensibly have one you know what to do 

LM399 is different since it has it's own oven and air draft protector built in (you can actually buy the chip without the white plastic cover). The only instrument I seen encapsulating the LM399 is the Valhalla 2720GS where 2 LM399 are inside an oven. They are probably running the 1st oven a bit cooler to compensate.

Even with it being ovenised there is still drift due to draughts on the can and legs, plenty of info on that and including the LTX1000. In your case I would be concerned that insulating rather than just stopping draughts will push the component temperature higher screwing with the reference circuit's voltage/calibration. You might find a more stable but out of calibration instrument is the result. Worth asking as a seperate thread in Metrology for some more experienced feedback.

[Kosmic]

I'm sure you'll agree though, from looking at the thermal image of the warmed up meter, that it's unlikely there is a stable thermal environment inside the case. Convection currents will roll around the inside constantly due to the uneven distribution of heat from the transformer. The fact that the reference is mounted high off the board on relatively long leads would suggest that local air currents could cause thermocouple effects.

Agreed. The temperature of the transfo is concerning. I'm wondering if it was always like that. This guys thought it was a good idea to add a heatsink and fan on top of the transfo.

https://www.pa4tim.nl/?p=2456

I don't see any noise specifications in the Solartron manuals, so I have nothing to judge what is acceptable for the 7061. I know that the 7075 is about twice as accurate when comparing specs, even though it only has 14 million counts compared to the 7061's 21 million counts. Even so, I was surprised when I saw how it behaved with the Weston cell.

I posted some comparaison on the Prema 5017 thread (Solartron 7061 vs Prema 5017 vs HP 34401a).

https://www.eevblog.com/forum/metrology/prema-5017-8017/msg2815402/#msg2815402
https://www.eevblog.com/forum/metrology/prema-5017-8017/msg2816952/#msg2816952

It may be that my suspicions are unfounded, but having the reference mounted high up off the board in a cool corner of a case with uneven internal temperature looks like a noise source to me. As you know, the 70[6,7,8]1 series meters all have an auto zero function which runs every 15 minutes to "correct for thermal drift" as the manual says. I think it's interesting that the older 7075 with its ovenised reference diode does not have this function and manages to deliver incredibly stable measurements to the microvolt level without it. Maybe there's some connection?

They definitely did a effort to move the diode away from the board for some reason. You can disable auto zero via gpib. I remember running some tests with auto disabled and did not see a huge difference. Technically it help with drift.

One other tool I'll be building to help look at the noise levels on all my precision meters is a xdevs.com KX LTZ1000A reference. I have all the parts now thanks to Mouser's incredible service. All I'm waiting for is the PCBs from OSHPark and some bismuth/tin super low melting point solder paste from China.

Cool, what is going to be your resistor setup ?