USA calibration club

[bitseeker]

Congrats on the new score, CM. I really like these old, fanless devices.

[Muxr]

I just scored a 34401a off of ebay so I'm now in the position to see if I can device some experiments to quantify the effect of this somewhat high internal temperature (and I'll have a reference to check the performance of any mods against)

Grats on the score! Looking forward to your measurements.

[cellularmitosis]

Hi, I've finally had a chance to look at the SVR-T measurements that I made last week.

Thanks for going to the effort to log these by hand!

[Conrad Hoffman]

The secret underground lair uses some elderly but useful equipment, like the big Tek scopes and the 1A7A high gain diff plug in. Here's a photo of the ref signal, AC coupled, 10 uV/div and 2 mS/div, with the filters set for a bandwidth of 1 kHz. I've got the ref in a shallow aluminum box that I can use as a guard. It took me a while to rid myself of some low frequency spikes/pulses. Not sure what's generating them in the house, but enough shielding got rid of them. Interestingly, my 731A has a lot more 60 Hz hum, but lower high frequency noise than the ref.

[cellularmitosis]

So cool!

[zhtoor]

The secret underground lair uses some elderly but useful equipment, like the big Tek scopes and the 1A7A high gain diff plug in. Here's a photo of the ref signal, AC coupled, 10 uV/div and 2 mS/div, with the filters set for a bandwidth of 1 kHz. I've got the ref in a shallow aluminum box that I can use as a guard. It took me a while to rid myself of some low frequency spikes/pulses. Not sure what's generating them in the house, but enough shielding got rid of them. Interestingly, my 731A has a lot more 60 Hz hum, but lower high frequency noise than the ref.

thanks.

how about a scope-shot at slow scan, say 2 Sec/div ?

regards.

[vindoline]

Nice scope shot Conrad! Now that I have a bit more lab space, I may keep an eye out for an old Tek with plugins. For a long time I've been interested in building a homemade diff amp and seeing what kind of performance I can get from the Rigol!

It looks like the first "upgrade" to the SVR-T might be a nice shielded enclosure to keep out RF and drafts.

[alm]

If you are short on space, getting a Tek AM502 differential amplifier plugin plus one of the smaller TM500/5000 mainframes might be an option. The circuit is probably very similar to the 1A7A (except solid state), it is basically a TM500 version of the 7A22. They have a BNC out that you can plug into any scope. Due to the smaller size and versatility they do tend to be more expensive. You might be able to pick up an 1A7A + mainframe for less than the plugin .

[Conrad Hoffman]

I was mostly looking for medium frequency popcorn noise, and didn't see any. Doing slow scans on the analog scope with a hand-held camera isn't practical. I'm getting ready to do data-logging and can see that time scale there, or if I have the time and the stuff still works, I have some paper chart recorders that would have been used for this sort of thing in the past. I'm hoping to be done with whatever I can do over the weekend and ship to the next destination on Monday.

I like the 1A7A because it fits my fat fingers well. The AM502 is definitely an option, and there are similar plug-ins for the 7000 mainframe series. The solid state diff amps should have better specs, particularly bandwidth, than the 1A7A. The 1A7A is also getting scarce and expensive when they turn up, and it uses Nuvistors. There is also the ADA400 for the newer scopes with power connectors around the BNCs, but IMO the thing is far less convenient than a dedicated plug-in. You could build your own version for very little.

[alm]

The downside of those older amplifiers with discrete components is drift due to temperature differentials. Only an issue if you want to do long term logging of noise for some reason. The ADA400 would be better in this regard. But more expensive and less convenient (need either a compatible Tektronix scope or buy/build a power supply like the Tek 1103).

I was mostly looking for medium frequency popcorn noise, and didn't see any. Doing slow scans on the analog scope with a hand-held camera isn't practical. I'm getting ready to do data-logging and can see that time scale there, or if I have the time and the stuff still works, I have some paper chart recorders that would have been used for this sort of thing in the past. I'm hoping to be done with whatever I can do over the weekend and ship to the next destination on Monday.

How about connecting a DSO to the Y output that the 500-series scope presumably has? Or does that sound to modern?

[Conrad Hoffman]

That would work. I hate to admit this in public, but I don't even own a DSO, though I could probably borrow one if necessary. And yes, one would certainly not use any of my antiquities for drift measurements.

[alm]

Guess another old-fashioned way to do this measurement is with a storage CRT, although I am not sure if there ever were any storage CRTs in the 500 series, never mind the space.

I meant drift of the offset of the differential amplifiers, so if you are trying to measure the noise over hours using some sort of external data recorder (trying to characterize noise of temperature, for example), the offset voltage may cause the signal to drift off screen unless you stop and readjust the offset/balance regularly. That is where more modern IC-based solutions like the ADA400 would be superior, since the temperature across that IC will be much more equal. I was not suggesting using these amplifiers to measure the ppm-level output drift of the references.

[Conrad Hoffman]

There was the 549. We had storage 5000 series where I worked long ago, and I've got a 350 MHz 7000 mainframe in the garage with storage, but old storage scopes are prone to CRT trouble, with no replacements to be had. If there's anything where a DSO is light years ahead, it's certainly storage and slow waveform work.

The big question is what time-frame are we interested in? I've seen popcorn noise and random jumps at pretty much any interval, but if you just stare at the scope for quite a few minutes, misbehavior should be obvious. Next step is longer term data collection at second or a few second intervals.

Drifting off-topic, many Fluke & HP gadgets used NiCad batteries for isolated operation and to keep the power up during a power failure. The 845 null meter and 731 references do it. Unfortunately, they had primitive charging methods and also used the batteries as a voltage clamp/regulator. I discovered the batteries in my "golden" 731A were shot, though they were still clamping the voltage to something reasonable. My other two units were noisy because the batteries were missing. Though I've made new battery packs in the past, they eventually suffer the same fates. NiMH batteries don't work well in this application because they can't take as heavy a trickle charge indefinitely. What seems to work well is to replace the batteries with a good sized capacitor, say 2000 uF or so, paralleled with a zener that clamps at about 14.5 volts. The combination provides the filtering and voltage limiting similar to the battery, though obviously not the remote power operation. I've never seen significant hysteresis when powering up the 731 references, so avoiding a power failure at all costs isn't very important, at least at ppm levels. Maybe different at tenth ppm levels.

Having fixed that, the SVT is only slightly noisier than my 731 references, and far less susceptible to hum, especially if powered from a regulated wall wart.

[bitseeker]

Conrad, what about a series resistor with the NiMH batteries to bring the current down to a safe level for their chemistry? I was considering doing that with my Keithley 228A, which has a very dead NiCd backup battery, though it's not used for clamping/filtering, just backup power.

[Conrad Hoffman]

Apparently NiMH can take a small trickle charge with little trouble, but I never see them used in float applications. Most things I'd need batteries for require 12 1/2AA size, with tabs. Too expensive and probably hard to pick a resistor that balances circuit requirements and trickle requirements. If I absolutely needed battery backup, I'd be more apt to use an external lead acid gel cell. Those too, have limited life. I used to have a bank of them for my standard cell oven, but disposed of that a few years ago.

[Muxr]

Speaking of batteries, a few days ago I ordered a set of VRLA batteries (lead acid but safe for office/lab use). http://www.ebay.com/itm/2-LOT-NEW-YUASA-12V-9ah-VRLA-UPS-Backup-Battery-Replaces-Werker-WKA12-7-5-F2-/191886670473?hash=item2cad563a89:g:EYQAAOSwe-FU455l

My intention was to use them as a power supply for noise measurements.

[Andreas]

Apparently NiMH can take a small trickle charge with little trouble, but I never see them used in float applications.

Hello,

after my experiences with my LTZ1000 references:
The limit is somewhere 20-40mA for AA cells.
Above around 60mA they get cooked.
For 1/2AA cells the values might be lower.

What I use is a L200 CC/CV regulator.
Charge current limited to C/10 (200mA) and voltage limited to around 1.45V/cell.
Since I use a diode to prevent discharging back. The actual voltage is around 18.1V.
And has to be adjusted so that the floating current does not get too high.
(max 60 mA including the 20mA for the LTZ1000 reference).
Unfortunately the T.C. of the cell voltage is negative so the whole works only for room temperature.

With best regards

Andreas

[Conrad Hoffman]

Finally, after getting my adult "make a living" stuff under control, I got back to the SVT. Wanting a less intrusive power supply than my big Kepco bench supplies, I found a 24 VDC wall wart (got a case of 'em) and made a nice little LM317 regulator board to for 15 VDC. Isolated, stable and quiet. I'll put that in the box when I ship it.

Original plan was to use the 845 null detector and record the output. It's surprisingly noisy and probably needs a major tune-up. Then it occurred to me that the 845 isn't needed. The little HP 3478A, that I already wrote a data logging program for, has a very sensitive lower range, 0.001 mV. Just fine as the null detector. For a first go I put the SVT back to back with my Fluke 731A and recorded about ten minutes worth of samples at one second intervals. Since the SVT is known to be nearly dead on from others measurements, I'll assume my 731A is about 2.5 ppm low. This agrees with measurements of both using my HP 3455A. I'll do a longer run tomorrow.

[cellularmitosis]

Great idea using the lower DMM range differentially against a more stable reference!

I just realized one could do the same to measure two LM399's against each other, or two LTZ1000's (assuming they match within the lowest range on your meter).

[sokoloff]

Alright; it looks like I can finally play.

I got one of my HP3456As working (no calibration), and have 3 Agilent 34401A (6.5 digit as well, cal expired Oct 1, 2016) that I just bought at the Blackberry/RIM auction. Also have an assortment of GPIB parts, but no proven GPIB data-logging solution yet. I hope to wrap the GPIB part in the next 4-8 weeks, depending on how complex it turns out to be.

[Conrad Hoffman]

I have a bit of cyclic wander over an hour or two, probably the dehumidifier turning on and off. Or something else TBD. The room temperature has been dead stable at 73F. Smallish, half a ppm or so. Right now I'm running some casual temperature tests with a cardboard box and desk lamp. Here's the datalogger screen I wrote for the HP3478A. It's reasonably functional, though by no means bug free. It's written in PowerBASIC for the Prologix GPIB unit. No graphing, just gets me a csv file for Excel.

[Conrad Hoffman]

This has been fun! I'll ponder the data for the next few days, but whatever temperature sensitivity there is, is darn small, buried in whatever other variations exist in my system. When we have a really good cal on the SVT, we can all adjust, but my guess is it's near dead on, and I'm running about 2.5 ppm low on my "golden" 731A.

I couldn't include the wall wart, as it would have put the package well over the limit. I did include the small regulator board with the mating DC connector, should anybody want to use it, and a small capacitance standard on a dual banana. Normally I'd do a shielded box with two BNCs, but few people would be able to take advantage of that. Instead, the cap is tagged with a three terminal measurement done on my GR 1615 that one can use with common bridges that measure independent of the front panel. It's a very good NP0 from Corning. I had wanted to throw in a resistance standard, but the weight was again too high, and I didn't have enough measurement history on it yet to feel comfortable sending it. Maybe next round.

I should be able to hit the post office tomorrow, so Dr. Diesel is next!

[cellularmitosis]

Ohh, more goodies to test!

Thanks so much Conrad!

[vindoline]

my guess is it's near dead on, and I'm running about 2.5 ppm low on my "golden" 731A.

Thanks Conrad! I'm just curious - why do you guess the SVR-T is "dead on" and your standard has drifted? It's always that chicken and egg question!

[Conrad Hoffman]

Somewhere back in this thread I thought the SVT got measured with a far better meter than I've got, and it was very close. Though I have three Fluke 731s that I can intercompare, it's been more than 13 years since the last traceable cal on my best one. I call that the "golden" unit because I modified it for near zero TC and it's the one that gets sent out. They're good, but I'd be astounded if they were that good. In the years since getting a cal, I've compared to various new Agilent 6.5 digit meters purchased at the day job. I'm always within a couple ppm, but I don't know how good those meter were when received- the chicken and egg problem!