Capacitor with negative ESR?

Discussion in 'General Electronics Chat' started by towheedm, Jun 17, 2013.

Thread Status:
Not open for further replies.
  1. towheedm

    Thread Starter New Member

    Jun 16, 2013
    Hello all

    I'm new to this forum. This is my first post and hoping I can get some help.

    I've been working on this piece of equipment with 8 capacitors in a bank. It's on the front end of the equipment and forms a voltage doubler network. There are 4 caps on the +ve rail and 4 on the -ve. It's powered from the utility mains of 208VAC-240VAC.

    The problem is that the equipment is exhibiting a rather large in-rush current with keeps popping the input fuse.

    After testing all of the caps with a Sencore LCR103 meter, I've found that all of the caps on the -ve Bus are showing a negative ESR on the meter.

    I've built a very simple model of the circuit and simulated it with LTSpice. The simulation confirms that with a negative ESR on the caps, the in-rush can indeed be extremely large. The simulation showed inrush in the range of teraAmps.

    These are Hitachi caps with the following specs:
    Cap Value: 8200uF
    Voltage: 400VDC
    ESR: 18mΩ
    Lser: 40nH
    Leakage: 0.01CV or 5mA (whichever is greater)

    When the caps are tested, the 4 on the -ve rail are showing ESR's of -0.05Ω, -0.26Ω, -0.15Ω and -0.35Ω. The caps on the +ve Bus are averaging 0.30Ω.

    I've googled on this but did not find anything. Any idea what a -ve ESR means and what effect it can have on circuits?

    Thanks in advance.
  2. t_n_k

    AAC Fanatic!

    Mar 6, 2009
    I very much doubt you have -ve ESR. It's most likely an artifact of the measurement on electrolytic capacitors. I also doubt the issue of fuses blowing is due to ESR . Is this something you are designing or existing equipment you are repairing? We also have to be careful giving advice on power supplies which are not transformer isolated.
  3. bountyhunter

    Well-Known Member

    Sep 7, 2009
    negative ESR is not possible
  4. towheedm

    Thread Starter New Member

    Jun 16, 2013
    Thanks for the speedy reply.

    This is an existing piece of equipment that I'm repairing. And yes, I agree that theoretically, -ve ESR should not exist. I've checked a known good set of the same caps and they all read around 0.3Ω.

    It's your standard voltage doubler circuit powered directly from 2 lines of the utility supply.

    Here's a simplified schematic of it. Initially, S1 is OFF. This allows the Bus to soft-start thru Rs. Once the Bus reaches a certain voltage ( about 460V). There is no inrush at this point and the initially start current is about 16A.

    S1 closes. The Bus then charges to about 660V. Measured peak current when S1 closes is roughly 200A for less than 1/4 cycle.

    S2 and S3 represents switching semiconductor devices. These are controlled by PWM signals from the control board. And this is where the problem is. Once the first switch turns on (I believe it's S3), there's a hugh -ve current pulse higher than 500A. At this point, the input fuse opens.

    In my simulation with LTSpice, the current levels seem to be expected, and was verified with a known good unit (except for the -ve current pulse) which I measured at just around 70A.

    And yes, I've already eliminated all other possibilities. The only unknown at this point are the caps. That's why I was as thorough as possible in testing them. The only discrepancy between my tests and the datasheet are the ESR.

    That's why I'm thinking a -ve ESR?
  5. crutschow


    Mar 14, 2008
    A negative ESR means the resistance is less than zero. This can only occur if the resistance supplies energy instead of absorbing it, in other words it needs an active amplifying device for that. That's not possible with a passive capacitor.
  6. towheedm

    Thread Starter New Member

    Jun 16, 2013
    Guess I'll have to see if there's any info on how the Sencore LCR103 actually performs it's ESR test.
  7. WBahn


    Mar 31, 2012
    Not surprising that a physically meaningless model produced physically meaningless simulation results.

    I suspect your model is missing other things that would be needed to get reasonable simulations of inrush current, particularly source impedance, wire impedance, wire inductance, etc., etc.

    It is almost certainly a measurement artifact of some kind, particularly if you are measuring components in-circuit. But even if you aren't, electrolytic caps have some strange behaviors that can confuse measurements if they aren't accounted for by the measurement equipment or, in some cases, by the person interpreting the measurements.
  8. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    It probably compares the capacitance value measured with a low frequency signal, to the capacitance value measured with a high frequency signal. If the cap has a lower capacitance when working at high frequency it has "high ESR".

    If the cap measures as a higher capacitance working at high frequency than it does working at a low frequency (Is that possible? I think so...) then it has a "negative ESR" by comparison.

    Of course what would really be happening is the real ESR is higher at low frequencies, and reduced at high frequencies.

    So no "negative ESR" actually exists but the cap is exhibiting a reversed relationship between frequency and ESR, compared to a normal cap.
  9. towheedm

    Thread Starter New Member

    Jun 16, 2013
    And that may explain the higher than normal current when S3 starts switching at approx 12kHz.

    This appears to be the case also with leaking electrolytics. I've tested several small value (10-470uF) leaky ones, and the LCR always indicates a negative ESR. I'm not saying that the ESR is actually negative, just that the LCR meter indicates this.

    I'm not sure exactly how LTSpice simulates a capacitor, just that the Help says it uses a special simulation matrix to simulate all of the params of the caps, ie: ESR (Rser), Equiv Serial Inductance (Lser), Leakage (Rpar) and Equiv parallel capacitance (Cpar).

    Simulating a simple series RC circuit powered by a DC source (and ignoring the effects of circuit lead resistance, inductance etc), show that if you set Rser to a negative value, the initial current (with the startup option set in the .tran directive) can in fact be quite high.
  10. Pane77

    New Member

    May 25, 2014
    Yes, I also have experienced negative ESR as well. I also was often wondering what that means!
  11. alfacliff

    Well-Known Member

    Dec 13, 2013
    discharge your caps for a while before doing the esr test. caps tht large sometimes build a charge back up after discharge. what your seeing as -esr is voltage across the caps. those sencores also will read strange esr if you test esr after checking leakage. just put a volt meter on the caps and you will see voltage.
  12. wayneh


    Sep 9, 2010
    Old thread, guys.
  13. bertus


    Apr 5, 2008
    Old thread, now dead.

    If you want to discuss the issue, start a new thread referencing to this one.
Thread Status:
Not open for further replies.