PSN-L Message - Subject: Re: Haunted system misbehaving again

From: Larry Conklin To: psnlist Sent: Fri, 22 Jun 2012 2:22 Subject: Re: Haunted system misbehaving again Well, I haven't tried locking the pendulum, but in an earlier=20 ncarnation of this problem (in '03!) it tried a few experiments where I=20 isconnected the lead that supplies the power to the oscillator and the=20 onnection to the feedback coil. Monitored the output and found that=20 ith the cables connected from the detectors I had significant noise and=20 ith the cables disconnected still pretty much the same. But if I=20 erminated the inputs with 2K resistors, the output went to zero, save=20 or a little +/- 1 count jitter. So, at that time I concluded that the=20 oard probably wasn't the source of the problem. Hi Larry,=20 This strongly suggests RF pickup. Can you tune a radio to=20 4 M Hz and find out what signals there are ? You are using two quite=20 sharply tuned circuits to detect movement. Do you have any RF=20 SCREENING around the seismometer ? It would be well worth trying ? s fate would have it, the thing has been running pretty well all day today. Snarl at your electronics occasionally ! On 6/21/2012 6:00 PM, Larry Cochrane wrote: Hi Larry, Have you tried locking down the pendulum so it can't move and letting=20 you system run for a while? Looking at the schematic I see two parts I=20 would replace. They are the two back-to-back 100uf (I think)=20 capacitors driving the feedback coil. I would replace them with one=20 47uf non-polarized cap. DigiKey has these=20 http://www.digikey.com/product-detail/en/ECE-A1VN470U/P1190-ND/227631=20 in-stock. . > Cold Solder Joint someplace ??? Chris Chapman made a similar suggestion, and I really can't rule it our=20 bsolutely. But Larry's bards are very well constructed and probably=20 ere wave soldered. So I don't think it's too likely. At one time I=20 ctually did try reheating a couple of connections that looked a little=20 uspicious, but it didn't make any difference. I had a Japanese TV set which started giving trouble in it's=20 power supply. I found that the solder blob on one component lead=20 was just not in electrical contact with the copper track underneath,=20 although it looked perfect and did not move when prodded. I=20 unsoldered it and was left with a shiny patch of solder. When I=20 scraped this off with a pen knife, there was a black insulating film=20 of corroded copper between the solder pad and the copper track. I=20 suspect that the board was not properly washed after it was plated.=20 Every single solder pad on the whole board was fraulty, but the=20 board still worked after the first pad was cleaned, washed and re=20 soldered. =20 > Im no expert, but Id not use electrolytics in any circuits but power=20 ircuits. The old fashioned "electrolytics" are pretty much a thing of the past. =20 think that good quality newer tantalums are fine. In any event, I=20 on't know what else you would use to get 47mf in a reasonable size=20 ackage. The Non Polarised electrolytics are quite good, but you do see=20 capacitance variations at very low frequencies. They also change=20 in capacitance as they age. Tantalums fail to short circuit, which=20 I don't like and are very sensitive to transient voltage overloads.=20 ..=20 Regards,=20 Chris Chapman . =20

= From: Larry Conklin <lconklin@............>
To: psnlist <psnlist@..............>
Sent: Fri, 22 Jun 2012 2:22
Subject: Re: Haunted system misbehaving again

Well, I haven't tried locking the pendulum, but in an earlier=20
incarnation of this problem (in '03!) it tried a few experiments where I=20
disconnected the lead that supplies the power to the oscillator and the=20
connection to the feedback coil.  Monitored the output and found that=20
with the cables connected from the detectors I had significant noise and=20
with the cables disconnected still pretty much the same.  But if I=20
terminated the inputs with 2K resistors, the output went to zero, save=20
for a little +/- 1 count jitter.  So, at that time I concluded that the=20
board probably wasn't the source of the problem.

Hi Larry,

    This strongly suggests RF pic=
kup. Can you tune a radio to

sharply tuned circuits to detect movement. Do you have any RF

SCREENING arou=
nd the seismometer ? It would be well worth trying ?


As fate would have it, the thing has been running pretty wel=
l all day today.

    Snarl at your electronics occasionally !



On 6/21/2012 6:00 PM, Larry Cochrane wrote:
> Hi Larry,
>
> Have you tried locking down the pendulum so it can't move and letting=
=20
> you system run for a while? Looking at the schematic I see two parts I=
=20
> would replace. They are the two back-to-back 100uf (I think)=20
> capacitors driving the feedback coil. I would replace them with one=20
> 47uf non-polarized cap. DigiKey has these=20
> http://www.digikey.com/pr=
oduct-detail/en/ECE-A1VN470U/P1190-ND/227631=20
> in-stock.
..

=
 > Cold Solder Joint someplace ???

Chris Chapman made a similar suggestion, and I really can't rule it our=20
absolutely.  But Larry's bards are very well constructed and probably=20
were wave soldered.  So I don't think it's too likely.  At one time I=20
actually did try reheating a couple of connections that looked a little=20
suspicious, but it didn't make any difference.

    I had a Japanese TV set which started giving trouble in =
it's

pow=
er supply. I found that the solder blob on one component lead

was just not in elect=
rical contact with the copper track underneath,

although it looked perfect and did = not move when prodded. I

<= /FONT> > Im no expert, but Id not use electrolytics in any circuits but power= =20 circuits. The old fashioned "electrolytics" are pretty much a thing of the past. =20 I think that good quality newer tantalums are fine. In any event, I=20 don't know what else you would use to get 47mf in a reasonable size=20 package.

PSN-L Email List Message