Renard's BMW

[BigMouse]

I've managed to get the file about consumption and efficiency displayed:

Nothing is showing up. I see the URLs when I click reply, but my browser can't load the images. Maybe try jpg files instead of png?

[woody]

They are TIFF images with a .png extension. I have converted them to png and re-uploaded here. Not sure it's what Renard wanted to upload, they look familiar...

[Renard]

Thanks very much Woody. My computer skills are very minimal.

The files -- one file split into two to be less than 200kB each -- diplayed OK when I uploaded them and when I checked them a few minutes later.

I did a screen grab of a pdf file which produces a tiff file, but by forcing a png extension on them they seemed to work. But later they disappeared.

How do you turn a tiff file into a png one?

To display mathematics, I use Latex which produces a pdf file, and then I have tried to get the file into a form which will display directly.

I really dislike the way maths displays in such an ugly and clunky way in common file formats, which is why I use Tex/Latex to produce much more elegant results.

[BigMouse]

In Windows, Alt+PrtSc copies the active window to the clipboard.
Paste it in Paint, crop as required, and save as png or jpg.

[Renard]

Further to my post of 27th August, the WS200 arrived back from Tritium yesterday with a new base, courtesy of James. He has described it as 'brazed, not bonded'. [Edit: not 'James Bonded'] I can only suppose that the original bases were susceptible to leaks at normal or even slight over-pressure of the coolant system.
I will also arrange for a coolant reservoir breather hole.

The coolant ports in the new base are at the other end from the original, which means I have to extend my coolant tubing. But It's probably a better arrangement as it reduces the busyness at the other end where all the cable entry ports are close together.

I haven't installed it yet, as I want to alter my HAZV box by replacing a contactor and introducing a DC/DC converter pre-charge, and it will be easier to do this with the WS out of the way.

[Renard]

I have been trying to think of a circuit to prevent a contactor closing until the voltage across its terminals is low.

Can people see any problems with this circuit?

[jonescg]

You might need to provide some context first, and then walk me through it. Which contactor is this? For the heater? For the traction pack? DC/DC is affecting precharge?

[Renard]

You might need to provide some context first, and then walk me through it. Which contactor is this? For the heater? For the traction pack? DC/DC is affecting precharge?

It's for the main contactor, which has been subject to a pre-charge. (But would apply to any pre-charge situation.)
On the left hand side, the 0V and 370V represent the contactor's main terminals; on the right hand side, the relay switched contacts are in series with the contactor's coil circuit.

The optocoupler is chosen to require only very small currents for the LED so as to keep the 450k resistors small -- with respect to their heat dissipation.

The diode in series with the 450k resistance is to protect the opto LED aginst reverse bias.

[4Springs]

This circuit looks interesting. Its purpose is sort of similar to one that I made to flash a light when above a certain voltage, but I like this one better. Let's see if I can understand it:

There is a relay in series with the traction pack contactor. The relay's normally closed terminals are in use, so that when we power the relay, we break the supply to the contactor.

The aim is that the contactor cannot be closed until the input on the left hand side (where it is written as 370V) gets up to a certain voltage. The relay is powered at low input voltage. Once the input reaches a certain voltage the circuit will remove power from the relay, allowing the traction contactor to close.

The 6N139 is an optocoupler. It will turn on when the current gets to a certain level. That is what the equations are up the top of the picture.
With a lower input voltage, and therefore input current, the opto is turned off. This lets pin 6 go up to 12V through the pull-up resistor. The LM308 op-amp input is taken high, which drives its output high. This turns on the transistor Q, which turns on the relay. The contactor cannot operate.
With a higher input voltage, and therefore current, the opto is turned on, which takes pin 6 to 0V. This is the input to the LM308 op-amp. So in this state, the output to the op-amp is also driven low. This in turn feeds the transistor Q, which is also driven off. The final result is that the relay is off, and the contactor is enabled.

So, for my questions and comments:

How much current is required to drive that optocoupler? I've not had much to do with optocouplers myself:
The equations at the top of the diagram state that at 29V (1.4mA) the opto is off, and at 370V (10-15mA) the opto is on. I'm not sure exactly when it switches though. I had a look at the datasheet (6N139), but couldn't work this out. I suppose a variable resistor in this part of the circuit could allow for adjustment.
If this was not accurate enough, I suppose that zener diodes could be used on the input to pass current at an exact voltage.

Why have the op-amp and the transistor? Wouldn't the transistor be sufficient?
Could there possibly be an optocoupler that would sink enough current to drive a relay? In that case the circuit could be simplified greatly.

If I have understood correctly, the relay is powered to disable the contactor. Wouldn't it be better to reverse the logic, and require power to enable the contactor?

Depending on the voltage chosen, and other circumstances in your vehicle, could it be possible that the traction voltage may fall below the threshold voltage while driving up a hill with low battery reserves? You don't want the circuit to cut out while driving. Perhaps make the relay latch itself on once enabled?

In my vehicle I rely on the driver to pre-charge. I have a circuit that displays a light on the dash for 30 seconds once the key is turned to ACC. I instruct any drivers that they need to wait until the light goes out before they then turn the key to ON. In your case I'd suggest an indicator to the driver to tell them that the pre-charge is happening now, and they shouldn't try to drive. Depending on the rest of your setup, it may be that depressing the accelerator during this time would discharge the capacitors, and make the pre-charge start again.

Thanks for sharing your circuit, it has made me think...

[BigMouse]

How much current is required to drive that optocoupler? I've not had much to do with optocouplers myself:
The equations at the top of the diagram state that at 29V (1.4mA) the opto is off, and at 370V (10-15mA) the opto is on. I'm not sure exactly when it switches though. I had a look at the datasheet (6N139), but couldn't work this out. I suppose a variable resistor in this part of the circuit could allow for adjustment.
If this was not accurate enough, I suppose that zener diodes could be used on the input to pass current at an exact voltage.

Why have the op-amp and the transistor? Wouldn't the transistor be sufficient?
Could there possibly be an optocoupler that would sink enough current to drive a relay? In that case the circuit could be simplified greatly.

If I have understood correctly, the relay is powered to disable the contactor. Wouldn't it be better to reverse the logic, and require power to enable the contactor?

The output of the optocoupler will rise as the LED current drop due to the internal pull-down transistor and external pull up resistor. At some point according to the published curves for the optocoupler, the non-inverting input of the op-amp will exceed the fixed 5.6v on the inverting input. The op-amp is wired as a comparitor.

As the voltage across the contactor terminals drop, the output of the op-amp will at some point transistion abruptly to "high", allowing the transistor to sink current and close the contactor.

The op-amp is there as a comparitor because the optocoupler does't "switch" discretely. It's a gradual increase. Renard is avoiding operation of the transistor in its linear range.

A tidy circuit. Nice work.

[Renard]

There is a relay in series with the traction pack contactor. The relay's normally closed terminals are in use, so that when we power the relay, we break the supply to the contactor.

The diagram is misleading; the Express schematic outlines only include a NC relay, but it should be NO, so that at 29V (the critical point) the relay closes and the contactor coil can operate.

The comparator -- or something equivalent such as a voltage follower -- is needed because without it, if the output of pin6 were used to drive the transistor base, the base current draw upsets the system and it doesn't work.

29V is the point at which the opto's output current -- that is, the current passing through its internal transistor between pins 6 and 5 -- is small enough to raise the pin6 voltage to 5.6V at which point the comparator output goes high. Of course this is adjustable either by changing the zener or by changing the value of the 4.7k pull-up resistor.

[coulomb]

Can people see any problems with this circuit?

The only issue I notice is that it will be quite sensitive to the characteristics of the opto coupler; they vary quite a bit between one unit and the next, and they change somewhat with age. I think their performance decreases by something like 10-20% over a decade or so.

This is not too much of a problem for you, since you are custom designing this circuit for your needs. The ageing issue is more serious though; it might nuisance trip in years to come, or silently no longer protect the contactor and capacitors from inrush (should one happen due to some other fault). If you get the thresholds right, it can still be doing its job even if the transfer coupling ratio declines by say 20%.

I suppose you could just check it every 5 years or so to make sure it's doing about what you want. A possible alternative would be to use a so-called "linear" op-amp, which use feedback. I think these might simply compensate for age by changing the feedback.

Opto couplers seem so easy to use, until you realise their limitations. Weber and I anguished over this since our BMU design relies on them heavily.

[4Springs]

The op-amp is there as a comparitor because the optocoupler does't "switch" discretely. It's a gradual increase. Renard is avoiding operation of the transistor in its linear range.

Ah, I see. I was treating the optocoupler as a discreet switch. So the level is set in the 12V section, not in the high voltage section like I was thinking. If I wanted an adjustable threshold then I'd make the 4.7k pullup resistor variable. That's why we need the 12V & 5.6V zeners (which was something else I was wondering about!).
Thanks.

[Renard]

Thanks for the feedback BigMouse and Coulomb.

About the ageing: a significant decline in the CTR over time would raise the turn-on voltage somewhat from 29V to something higher.
One should set this level low enough that this would not be a problem. Perhaps one could start at say 5% of full voltage (18V in this case) and accept the extra 100ms or so of pre-charge time required.

[jonescg]

OK, so this means that there is always going to be a 0.83 mA draw from your pack, unless any additional HV isolation contactors are open?

Otherwise it looks pretty neat.

[Shirker]

On the other side of the build vs. buy decision, you could send $55 to Ian for a ZEVA Smart Precharger, and get a couple of additional features... (I have not used one)

[Renard]

On the other side of the build vs. buy decision, you could send $55 to Ian for a ZEVA Smart Precharger, and get a couple of additional features... (I have not used one)

I see that it's voltage-limited to 320V. Without seeing the circuit I couldn't say if if could be changed to cope with higher voltages.

My simple circuit would set you back all of $5.

And Chris, yes, if all your pack contactors were closed except for the main one, you'd be using 0.8mA. But you could go further down the curve and use even greater R than 450k and so lower opto LED current. But better to work in the expected range as shown in the data sheet Fig 14.

[jonescg]

The Zeva precharger utilises a solid state relay for switching precharge. It's tough to get them rated for voltages much higher than that. However your circuit above would avoid that problem.

I am interested in finding a better solution to managing precharge and discharge for my bike. A few complications include the 700 V bus and the absolute requirement for discharging of the inverter capacitors to <50 V within a few seconds. I currently have a high voltage DPST relay switching the precharge and the closing of the main contactor is at the mercy of a timer. If I decide to turn the key mid-way through precharge (that is, about 500 ms after turning it on) there is the risk of a large arc developing inside the relay as it attempts to discharge the capacitors. While the relay is supposed to be able to handle the hot on-switching, its not rated for hot off-switching.

Also, should there be a large short in the 12 V system, a large voltage drop will occur, potentially opening the contactors under load. This would be another fail-unsafe arrangement.

SO a fail safe is needed, and some kind of voltage detection based system might work well. HVDC is a bugger eh?

[Renard]

So Chris, I see there are three situations.
1. Pre-charge
2. Break under load
3. Discharge

The first is covered by my circuit.
The second is the kind of major fault that happens once in a blue moon and which the contactor is designed to handle -- at least a few times.
The third is not as serious as it sounds because you say you have 5s to go from 700V to 50V. I calculate you can have RC as high as 68. (What value is the capacitance of the controller?) So R can be large which means that any device that switches R to controller neg. is minimally stressed.

[jonescg]

The way I currently have it, I have a 50 W, 520 ohm precharge resistor, impeding inrush current to the 560 uF worth of capacitors in the inverter. This means precharge is over in 1 second, and the main contactor closes after 2.4 seconds.

On turning the bike off, all contactors and the HV relay are opened, leaving the only closed circuit being the discharge of the capacitors (which takes about one second). In both cases, there should be no current flowing other than the 1.4 amps through the precharge/discharge resistor.

It's the opening under load scenario that I am cautious of. Say I am hammering along at 250 amps, 640 V, and for some reason my 12 V supply fails momentarily. All three isolation contactors and the main contactor will open, and the HV relay will switch to NC, potentially allowing an arc to be carried across the last contactor to open. I don't know if this will really happen, but I have seen some catastrophic events involving precharge relays.

If you like we can continue this in my build thread - viewtopic.php?title=voltronevo-jonescgs ... 553#p42850

[zeva]

The Zeva precharger utilises a solid state relay for switching precharge. It's tough to get them rated for voltages much higher than that. However your circuit above would avoid that problem.

Hi all The CPC1968 I use in the higher voltage precharger is actually rated for 500VDC, so not much good for Chris's 700V race bike, but would probably be safe working up to say 400V nominal packs. It'd probably just need higher value power resistors and voltage sense resistor.

The good thing about using an SSR is that it can safely break DC under load without any arcing (e.g if you have some error detection or precharge timeout). They are expensive components though - that SSR in the precharger costs more than the rest of the components combined!

[jonescg]

Thanks for popping in Ian. I knew they could take some high voltages but couldn't recall how much. I did know that it wasn't 700

If not for the race rule requiring discharge, I would just avoid it altogether as it's simply one more point of failure. There's no harm done in leaving caps charged up for a few weeks.

[Renard]

[quote="jonescg"] The way I currently have it, I have a 50 W, 520 ohm precharge resistor, impeding inrush current to the 560 uF worth of capacitors in the inverter. This means precharge is over in 1 second, and the main contactor closes after 2.4 seconds.

On turning the bike off, all contactors and the HV relay are opened, leaving the only closed circuit being the discharge of the capacitors (which takes about one second). In both cases, there should be no current flowing other than the 1.4 amps through the precharge/discharge resistor.

It's the opening under load scenario that I am cautious of. Say I am hammering along at 250 amps, 640 V, and for some reason my 12 V supply fails momentarily. All three isolation contactors and the main contactor will open, and the HV relay will switch to NC, potentially allowing an arc to be carried across the last contactor to open. I don't know if this will really happen, but I have seen some catastrophic events involving precharge relays.

If you like we can continue this in my build thread - viewtopic.php?title=voltronevo-jonescgs ... 553#p42850]

Yes, that's a good move. More over there.

[Renard]

At more or less the same time as the WS base coolant leak, I also suffered a breakdown at the local market. I had a neighbour tow me the 10km home, and then sent the WS off to Tritium. This was in late August. Then, with the WS base replaced, I still had the breakdown fault to deal with.

Meanwhile I have been engaged in the process of selling our house. Although we've been here 18 years, there were still some major matters to fix up and I let the BMW just sit while I organised the carpentry and painting and a host of minor repairs and touch-ups.
But now the house has sold and we have to move in early Feb. so it was time to get the fault diagnosed and repaired.
I sent a WS log file to James, and he suggested I check the encoder. Alas, it was sending no signal to the WS.

Nothing for it but to dig down to inspect it. A solid day's work and I hoisted out the three front battery boxes.
The photo shows the problem. The encoder lead must have been fouling the fan and twisted. Further proof of Murphy's Law and the importance of build quality at every point.




The effort is not entirely wasted however, as I have been planning to install a bigger fan for the motor which has been getting hot going up hills.

This is not surprising, since 1600kg up a 1 in 10 slope at 15m/s (54km/hr) requires 23kW at the wheels, or about 24kW of mechanical energy -- about 33kW electrical -- at the motor, just to do the lifting. Additionally of course, 54km/hr horizontally requires about another 5 - 6 kW electrical, so the motor puts out about three kilowatts of heat.

I'll break this post into two.

[Renard]

The existing fan, recommended by Tritium, is a powerful little thing for a 120x120 unit. Its performance curve has zero volume at 350Pa and maximum volume of 355m^3/hr at zero pressure, with an inverse, approximately linear response of pressure and volume between these limits.

I had been considering adding another fan in series which could be bolted on the outside of the fan cowl as a direct input to the existing fan. The effectiveness of such an addition depends on the point on the response curve where the existing fan is operating. It would be less effective if that point is towards the high volume/low pressure end, and most effective at the mid-point of the curve.

To find this out, I pushed a tube of water into the fan-cowl/body gap and measured the water column height difference: it was 7 - 8mm indicating a fan air pressure of about 75Pa. From the fan performance curve, this indicates a volume flow of 300m^3/hr. Since the maximum volume at zero pressure is 355m^3/hr, there didn't seem to be much advantage in adding an extra fan.

The other strategy I considered is to extend the cowling further down the motor body. But this is not possible on the base, nor on the terminal box side, so it could only cover two sides. And I wonder if it would help much anyway.

The other course of action that looked the best was to fit a larger and more powerful fan. So I have ordered a 200mm diameter circular one that seems suitable. The fan has 3 1/2 times the capacity of the existing one, but runs from 48V. That calls for another DC/DC converter, albeit one with a smaller output than the existing 12V, 450W Meanwell, for example a 48V, 150W type.

I don't have much room left under the bonnet for another piece of equipment, so I'm putting it in the boot. Fortunately I had run two surplus wires in the same conduit that encloses the charging link that connects front and rear so they can be used to supply HAZV to the new converter.