Bugzilla – Bug 3872
110V relay constantly toggeling
Last modified: 2006-10-11 14:21:36 UTC
From time to time Transporter would make clicking noises so I tried to figure out why. I realized it's the 110V relay. There seems to be a problem in the automatic voltage adjustment, probably caused by the wrong measurement of the input voltage and / or it needs a hysteresis. Here is what I see: Input voltage Current Settings 110V Relay (measured with voltmeter) AC line voltage 256 VAC 189 VAC off 244 VAC 176 VAC off 232 VAC 163 VAC off 220 VAC 150 VAC off 209 VAC 137 VAC off 196 VAC 135 / 124 VAC toggeling @ about 2Hz 182 VAC 121 / ??? VAC toggeling @ about 5Hz 171 VAC 128 VAC on 160 VAC 118 VAC on 150 VAC 105 VAC on 140 VAC 93 VAC on 127 VAC 80 VAC on 115 VAC 67 VAC on
yep there is a scaling factor that needs to be fixed feel free to have a go at it - sounds like you have a variac to test with?
It a kind of variac, but it only allows steps of about 10VAC. I played with the scaling factor (it is currently set to 62) and so far the best value is 46. Unfortunately it's not linear. How exact would you like to have the measurement? With this value (46) the maximum error is about 16% for some voltages. There is another problem I do not know how to solve (or if it needs to be solved at all). If the measured voltage is higher than (currently) 250VAC the power is turned off. Which leads to some lower measurement which turns the power on again, but since it's to high the power is turned off again and so on.
The oscillating over 250V is a bug - it should turn off and stay off. As for the non-linearity - we could correct this with a lookup table unless there is a simple formula that fits. I don't think it's a big deal though - the main thing is getting the right relays to activate for a given input voltage, not the exact absolute readout.
I tried to use two different formulas for 110 and 220 VAC and also just one scaling factor that would give readouts closer to the actual voltage, but without success. With the current scaling factor the readout is much too low, but whenever I try to fix that I end up with the following problem: The sleep relay is not activated if I power up with about 220 VAC. Also trying to fix the over 250 VAC issue wasn't successful so far. I figured the easiest way would be to turn off the sleep relay and not to call the voltage monitoring function anymore. This works as intended, but unfortunately it also turns the sleep relay off for voltages below 250 VAC.
If the new scaling makes the voltage readout accurate, then perhaps you just need to change the switchover thresholds now?
Ok, I now use two different scales for 110VAC and 220VAC mode which give me accurate readouts (within a few volts difference). I also upped the switchover threshold from 120VAC to 165VAC (as you suggested). The over 250VAC issue looks like this: Whenever the power is turned on it takes about 2.5 seconds until the voltage readout is accurate! Unfortunately during this 2.5 seconds the voltage readout is about 2000 too high which translates into about 30VAC which causes the sleep relay to turn off at power on for an input voltage of about 230VAC (230 + 30 > 250). Some ideas to solve that: - Wait about 3 seconds after power up before checking for the first time - Increase the maximum allowed input voltage (currently 250VAC) - Find a way to make the voltage readout more accurate right after power up (don't know if this is even possible)
hmmm - i didn't realize it takes that long to stabilize. It's not critical that the relays come on extremely fast. It would be okay to wait longer to trigger the second relay (and then start the audio thread). I'll have to test the behavior with your changes and watch it on the scope to get a better idea...
I am now having problems with the latest firmware build, as the ac power shuts off! My multimeter currently reads my power between 239-241 VAC, but the firmware is reporting 261-262 VAC. I do not see any delay in the voltage reading stabilizing on power up. Sean have you had change to look at this yet? Felix might the issue be caused by your variac stabilizing when load is applied? At the moment this issue will block me from releasing another firmware image for testing.
Richard - I've changed the shutoff voltage from 250VAC to 270VAC so you should be ok for the moment. Sean - Do you think that's save or should we go back to the old code for the time being? Sean - I've asked everyone to measure the main voltage with a multimeter and to check the readout on Transporter. My current feeling is, that we have different readouts (for the same input voltage) on different units. Do you think that's possible? Anyway I found another issue with my variac. Using the new code and going through the variac I get readouts that are within 5VAC of what I measure with the multimeter, but if I connect Transporter directly to mains I get a lower readout (i.e. 232VAC mains -> 215VAC readout). And to top this, Richard gets a _higher_ readout with the new code (i.e. 240VAC mains -> 262VAC readout).
Sean - what do you think about that? I do not have much data yet, but it looks to me that at least the readout on Richards unit is faulty/different? Some data with the old code: Mains Readout Difference Reporter 232 138 -94 Felix 239 147 -92 Adrian 240 188 -52 Richard (that's about a 40V difference) 122 91 -31 Chris 122 92 -30 Chris Some data with the new code: Mains Readout Difference Reporter 232 215 -17 Felix (this is without variac) 240 262 +22 Richard (that's about a 39V difference, so at least this is consistent)
Created attachment 1471 [details] Voltage measurements Voltage measurements from Chris on a v2.0 mainboard
The differences between Felix and my pre-production units were due to the wrong version of the digital input board being installed in Felix's transporter. This also caused the delay in getting an accurate reading at power on. The code has been adjusted and now works at voltages around 110V and 230V.
(In reply to comment #11) > Created an attachment (id=1471) [edit] > Voltage measurements > > Voltage measurements from Chris on a v2.0 mainboard > I expect there to be some "dead zone" between the upper end of the "120V" range and the lower end of the "240" range. However, you did not indicate where it comes back on... at some point as you increased the voltage, the red lights came back on, right? By the way, the red lights are actually a functional part of the super regualtor circuit. They are not indicator LEDs. The correct test for an acceptable input voltage is to see if the voltage _before_ the super regulators is at least 20VDC for the +/-15 rails, and 10VDC for the 5VDC rail. Lower than that, and the regulators may still output sufficient voltage but it will not be as clean. Mike can show you how to measure this.