HP 3458A on a "Pure Sinewave" UPS

[Tony_G]

Hey All,

I'm just wondering if there is any issue with putting my 3458A on a UPS? I wanted to keep the PC that I'm using to record data from my 732A up if a power outage occurs and I'm wondering if I should just keep the 3458A on it as well?

Does anyone have experience with doing this?

Thanks,

TonyG

[KT88]

Since UPSs use  switched power stage s it won‘t be the cleanest supply voltage. Although you could use it detached from mains there is still some HF common mode noise to ne expected.

[bdunham7]

The only issue I've had with my APC SMT1500 UPS is that the small amount of conducted noise that it has interfered with the noise floor on my Fluke 8506A in the lowest AC ranges.  I now have it set up so that I can take it out of the picture (which requires both a shut down and unplugging) reasonably easily.  I haven't seen any effects with 10V references or the meters measuring them.

[johansen]

If its a problem then a 100 watt isolation transformer will fix it.

[J-R]

Some UPS units like the APC SMT1500 (I have a few of the Dell/EMC-branded ones) have a "Green" mode which can be enabled or disabled.  When enabled, the UPS is essentially in bypass mode with most of the electronics powered down and you are connected directly to the wall until there is an event.  When disabled, it operates like a normal double-conversion or AVR or etc. UPS where it essentially is running the inverter all the time and there is no direct connection to the wall, in a manner of speaking (ground is always connected of course).  Also, there is a slightly longer delay before the UPS intervenes when it is in green mode due to the relay switch delay.

I've had most of my equipment on pure sine UPSs for years (in green mode) and haven't ever had any issues, but I can't say I've run any specific tests to see what things look like when the UPS is supplying the power or with green mode disabled.  It probably depends on multiple factors, so running some tests would probably be the best way to determine which is better for you.  I wonder if some filtering could help as well.

[sweaterfluid]

UPSs employ switched power stages, therefore the supply voltage will be less clean. Although it may be used independently of the mains, some HF common mode noise is to be expected.
hill climb racing

[IanJ]

My tuppence worth…….
If you have a UPS that isn’t producing the best sine wave output then look at adding an isolation transformer and an active or passive filter/ line conditioner. You could probably build something quite small if it’s only powering a couple of instruments.

Hmm, might have a go at this myself on my own UPS.

Ian

[jfet]

Many, many years ago, I was involved in testing LM199A voltage references.  We used saturated mercury cells in series and measured the LM199 differential using an Fluke 845 (? been a long time).  The 845 had a chart recorder on it. 

We had the whole setup on a 1KW sinewave UPS.  We learned early on, we could not afford to lose the test because of power outages.  After the UPS was added, I do not remember any issues with the data. 

I would think a HP3458A would run fine on a sine wave UPS.  Should be easy to test, bu pulling the plug and see what happens.  Maybe check the unit with the input shorted and then with low noise/battery source on input close to where you plan to measure.

[IanJ]

I built this.......to try with my UPS/3458A.

Isolation transformer (625VA) and two filters (L 1.0mH, Cx 0.75uF, Cy 0.01uF), one before and one after the toroidal.
Incoming earth and outgoing earth are connected.
At bottom of pic is just a regular UK double 13A socket for the output. Overall, good for maybe 3A so more than enough.

Bought from Ebay, added the filters myself. Typical cheap made the paint wasn’t scraped back on the incoming earth stud and was relying on a star washer poking through the paint!

Not happy with the toroidal fixing, needs layer of thick neoprene under steel bar I think.

Testing soon…..

Ian.

[dietert1]

I made a synthetic mains supply in order to produce a more stable voltage. The meter was running inside a thermal chamber and i didn't want temperature changes caused by mains voltage variation. My setup had a 500 W class B audio amplifier and a 180 W transformer to make 230 Vrms from 24 Vrms. The amplifier got a 50 Hz sine from a DDS type generator. This worked very well. One could add synchronization with real mains in order to reduce hum.

Regards, Dieter

[EC8010]

A couple of points about the toroidal isolation transformer:

(1). If I'd seen that at work, I'd have outlawed it. An isolation transformer should only have one outlet. The reasoning: Suppose person A works on a live chassis powered from the isolation transformer. That's fine - it's what an isolation transformer is for. Now suppose person B works on a live chassis (connected the other way around) plugged into the second outlet of the isolation transformer. Person A hands person B a screwdriver and both are electrocuted.

If you connect neutral at the outlet to neutral at the inlet and treat the thing as a sophisticated mains filter rather than an isolation transformer, there's no problem.

(2) Toroidal transformers have much lower leakage inductance than EI transformers, so they have an excellent high frequency response (but high shunt capacitance). I couldn't see if it has an E/S screen (quite tricky to do on a toroid), but a toroid's high interface area between primary and secondary means high capacitance between primary and secondary, coupling edges. In short, both differential mode and common mode filtering isn't as good as you'd get from an EI isolating transformer having a foil E/S screen.

Still, even if it proves disappointing as a filter, it probably has other uses.

Sorry to rain on your firework.

[IanJ]

Just trying to set up some initial tests, playing around with WinGPIB, DHO914 and Micsig DP10013 100MHz HV diff. probe.
Measuring VAC RMS isn't going to give me much in the way of noise analysis other than those that spike above or below the Max and Min values, and of course visually on the graph. But anyways, some results below.

Three graphs:
Across L& N, measuring AC RMS
GPIB command = :MEAS:STAT:ITEM? CURR,ACRM,CHAN1
200 samples, sample rate 0.5sec.
Micsig set to X500 attenuation
No load on output.

Note the Incoming AC and outgoing AC scales on the graphs have been set the same, the "shorted" graph is different.

1. Probes shorted together
Graph Max-Min = 0.000089 = 0.0445 VAC

2. Incoming AC (probes across L & N)
Graph Max-Min = 0.003579 = 1.7895 VAC

3. Outgoing AC (probes across L & N)
Graph Max-Min = 0.002360 = 1.18 VAC

So, the max VRMS deviation of the output is less than at the input......however, need to work out what I can measure on the DHO914 that will record better the noise levels.

Shorted:


Incoming AC


Outgoing AC



UPDATE:
Another two graphs, VAC RMS, 2400 data points, 40mins, 1sec sample rate.

Incoming AC



Outgoing AC

[EC8010]

Mains interference is (fairly) random. Why not explicitly test your transformer/filter with sine waves from a function generator?

For common mode: Short the primary winding, short the secondary winding. Terminate RG58 cable from generator in a through termination. Connect centre BNC pin to shorted primary, outer to chassis. RG58 to oscilloscope, oscilloscope set to terminate input. Centre pin to shorted secondary, outer to chassis.

For differential mode: Terminated signal across unshorted primary. Input of oscilloscope across secondary. Decide whether termination makes more sense.

In both cases, measure spot frequencies between 100kHz and 10MHz. We're talking low RF, here.

But what might a UPS produce that could upset an elderly precision DMM? I wouldn't be in the least surprised to learn that a 3458A has a linear supply using an EI mains transformer. If that's the case, what will upset that transformer is DC. A trapezoid is effectively switched DC and will cause saturation of the transformer's core, leading to excess leakage flux, and that might creep into measurements. The isolating transformer is a high-pass filter, so it might assist. And it might not. A simple oscilloscope trace should tell you what's going on. You don't want to see flat tops on the mains. And yes, I have noted that the thread title includes "Pure sine wave". How pure is pure? I'll bet that a UPS manufacturer's idea of pure is different to mine.

[IanJ]

Some AC RMS tests on the actual UPS output tied into my transformer/Filter.
Note UPS in pass-thru, not actually running off batteries....thats to come.
UPS is a Compaq T2400H.

1. From UPS - Transformer Incoming AC (probes across L & N)
Graph Max-Min = 0.01046 =  5.23 VAC

2. From UPS - Transformer Outgoing AC (probes across L & N)
Graph Max-Min = 0.00706 = 3.53 VAC

So, the AC RMS deviation figures from the UPS are higher than the non-UPS AC supply per my last post, but the transformer/filter does seem to be helping.
More work required.

Incoming AC (into transformer/filter):


Outgoing AC (from transformer/filter):


Ian.

[kbrill]

The older commonly available used at least in the larger 3000 W and larger sizes may be the right choice .

Combining a Ferroresonant  power supply for voltage stabilization and filtering , it only goes into a battery backed up switching mode during significant voltage surges , sags and interruptions .

Eaton's Ferrups power suppliers are a continuance of the product line of Best Ferrups , formerly Best Power .

Extremely well made and durable .


Eaton has discontinued this product line , however used it may be the correct choice

https://www.eaton.com/us/en-us/catalog/backup-power-ups-surge-it-power-distribution/eaton-ferrups.htm

[coromonadalix]

why not adding an emi rfi filter on the ups  ac output ?  it would help ?  no ?