Magnetic field surrounding battery

[raithrovers1]

A technician of mine was performing battery maintenance on a UPS battery containing 2 strings of VRLA batteries. Each string had 40 jars, 240 Cells and were being float charged at approximately 530VDC. The battery is connected to a commercial type UPS (Powerware 9355) which has no isolation transformer on the input. Therefore there was high AC voltage between the posts of the battery and the ground at approximately 237VAC. (Very dangerous in my opinion but that is another topic!)
They went to take the AC ripple current going into the battery from the rectifier in the UPS. They do this by using a Fluke i410 amp clamp connected to a Fluke 87 meter clamped around the main battery positive cable. They found that they were getting very unusual readings of between 10AAC to 40AAC.
What was more unusual was that when they disconnected the amp clamp from the cable the reading on the Fluke went up to approximately 150AAC! This confused them. They proceeded to move the amp clamp all around the battery (as if it were an antenna) without connecting around any cables. At some points the meter read upwards of 300AAC! the same affect could be found when the meter was changed to DC. If you moved the meter/amp clamp away from the battery the spurious readings disappeared. The amp clamp is calibrated and is working as per design.
What is also interesting is that when internal resistance on this battery was taken using an Alber cellcorder - the cellcorder stops working. This battery has caused 3 cellcorders to be sent back for repair!
It should be noted that there is another battery in the same room as this battery that is being charged by a regular switchgear charger. This battery does not show the same affect.
Has anyone seen a phenomenon like this before?

UPS Service Manager

http://www.nolanpower.com

 

[VEBill]

"Each string had 40 jars, 240 Cells and were being float charged at approximately 530VDC." ...and... "What is also interesting is that when internal resistance on this battery was taken using an Alber cellcorder - the cellcorder stops working. This battery has caused 3 cellcorders to be sent back for repair!"

How does the Alber cellcorder connect to the fairly high voltage bank of batteries?

 

[dpc]

You'll get a current reading when the CT is moving. I'd suggest a scope to see how much ripple voltage there really is.

 

[sibeen]

Just looked at the spec sheet for a Powerware 9355. It is apparent that this unit uses an active front end rectifier and uses what Eaton calls Advanced Battery management (ABM) technology. From this I can surmise that the battery is not connected directly to the rectifier bus. A separate charger board will provide the charging current to the battery bank and ABM lets the trickle charge to the battery be disabled. The batteries will then be connected to the rectifier bus through some blocking diodes, or perhaps SCR.

So this unit is going to have a separate battery charger which is connected directly to the battery and will utilise some form of high speed PWM switching to develop the DC voltage. Normally the charger will just be a card within the UPS as you're only trying to generate 10 amps or so of charging current - depending upon the size of the UPS. Now the output of these style of chargers is DC, but it also has high speed switching artifacts (20 kHz?) imposed on the DC. The Fluke i410 has a usable frequency up to 3 kHz, so it's not going to be able to accurately read higher frequencies than that, and in my experience was always crappy at reading anything that had switching transients imposed - sorry Fluke fanboys

Another thing to note is that your battery voltage was sitting at around 530 volts. This is about 2.21 volts/cell which is very low for a charging voltage and would indicate to me that the UPS was in ABM mode meaning the charger was actually turned off at the time of the readings, or the battery was actually discharging at the time. Where your people performing a discharge test perchance?

And finally, as to the input transformer, realistically, except in exceptional circumstances, no UPS of this rating (up to 40 kVA) has had an input transformer included for nigh on 20+ years. At this rating they are all fully active front ends, or a hybrid with a rectifier followed by a boost section.

 

[raithrovers1]

"You'll get a current reading when the CT is moving. I'd suggest a scope to see how much ripple voltage there really is. "

The amp clamp did not have to be moving to pick up the field. If it was stationary next to the battery the reading was still very high.
They did attach a scope (Fluke 190) which was seeing the same effects. They cant remember the frequency of the ripple but is was above 20kHZ.
What was also interesting is that when the amp clamp was removed from the scope and the scope was placed near to the battery it also picked up the field.

I also spoke to the tech to get the correct info from his report. He had mixed up which type of UPS it was (as there were several models at this site). It was actually a Powerware 9390 which still has an active front end and ABM.

Also, I asked the tech what the actual voltage float voltage was and he said it was 545VDC. The previous voltage I had mentioned was from his memory. 545VDC was from his report.

"And finally, as to the input transformer, realistically, except in exceptional circumstances, no UPS of this rating (up to 40 kVA) has had an input transformer included for nigh on 20+ years. At this rating they are all fully active front ends, or a hybrid with a rectifier followed by a boost section. "
Well I would agree if this was a commercial environment but we work in petrochemical plants and the majority of UPS's that we work with are industrial which all have input and output transformers and a floating DC bus. Even as low as 10KVA.

Thanks again for your replies.

UPS Service Manager

http://www.nolanpower.com

 

[ScottyUK]

A possibly related problem (or possibly not) which I've seen with some of the older Hall-effect transducers made by LEM is that the current output signal can be modulated by a voltage on the conductor. I've seen this on a few occasions, some in medium-frequency inverter applications where high dv/dt occured during commutation and once on a DC conductor prone to sudden shifts in potential relative to earth. In all cases the solution was to provide an earthed electrostatic screen between the conductor and the transducer, which suggests that the problem is capacitively coupled rather than magnetically coupled. If you have a lot of HF switching hash on the battery leads then I wonder if you are seeing something similar? If you want to test the theory then wrap a piece of earthed foil fully around an insulated section of conductor and place the current probe over the section with the earthed foil. Obviously be careful when you're doing this and keep away from anything energised.

Aside: does your organisation have any experience with AEG's Protect 8 series of industrial UPS's?

 

[DanEE]

Saw similar problems on the Powerware Plus UPS system in that the Alber Cellcorder would not work on that system only.. The same Alber worked fine on rather large variety of other type UPS systems that we maintained. I suspect the problem is the nature of the noise as suggested in the above post, (possibility of pulse charging being used?) on the battery string. As we know, if the noise or charging current is non sinusoidal‎ in nature, any instrument other than a True RMS reading device is going to give erroneous AC readings, and I would think sufficient AC noise present during the attempt to read DC current with a hall effect device would likely be erroneous.

Would be interesting to connect a current probe (with sufficient high frequency response) to an oscilloscope and see what type of current waveforms are on the battery string that you are having a problem with.

I got interested in pulse charger technology some time back and read a number of the patents on the subject and did a little measuring on my own.

On one small pulse charger I measured, I used a high frequency "air core" current probe and an oscilloscope and found a ringing pulse (fundamental frequency around 3mhz) on the battery charge leads, being applied at a fairly fast interval rate.. not a suprising find given the methodology described in the patents below. As a follow up, I took some old, pretty bad condition AGM batteries off a couple of sites that had charger problems (i.e. sulfated batteries) to test if pulse charging really worked. Some of the batteries were just too far gone (and would go open during bulk charge phase), but others that successfully bulk charged, and then went on pulse charge recovered to useable capacity quite well. Made me a believer of pulse charging systems. Although I don't know the detail, I've seen in some of the commercial scale APC systems, that they are using more advanced charging methodology and claim longer battery life as a result.

https://www.google.com/patents/US82...a=X&ei=r0EsU9ypOISo0gHD24GoAQ&ved=0CDYQ6AEwAA

https://www.google.com/patents/US6677730

http://www.pulsetech.net/content/FAQ.aspx

 

[raithrovers1]

Scotty: I have no experience with the Protect 8. Only with the older industrial systems such as the transobloc and a little knowledge on the protect 3.
As a note, I haven't managed to get someone back to the site to look into the problem further. I do have someone going within the next two weeks so I will hopefully have more info for you then.
Thanks as always for all the comments so far.

UPS Service Manager

http://www.nolanpower.com