Rich Weyand's CoolerCrawler analog throttle 2025 update

[diesel]

I have started work on re-creating a very close rendition of Rich Weyand's CoolerCrawler from the early 1990's (or maybe late 1980's). It was said to have superior performance compared with other systems on the market, which Rich said, he found important with the N scale locomotives he was running. I have made a circuit board and can't get the parts until next week... I can test in one direction only until the relay arrives (likely 2 weeks). I will continue with updates here as the project moves along.

 

[Kusojiji]

Very nice that all the component numbers are on the board.

 

[gasnbrass]

Interesting to see that the old cooler crawler throttle is still popular. If I'm reading the PCB correctly you've made a few changes from the original. Notably some mosfets and the momentum cap looks to be 1/10th of the original, but I suppose that's due to the mosfet doesn't use hardly any current to operate. I have been testing this circuit myself and overall I really like it and think it performs well but noted several things:

1) I don't understand the function of the final 1ohm 5w resistor and zenier. I've read it described as a feedback sense resistor (but I see no feedback loop) or as part of the zenier circuit. I don't see how the zenier is "venting" any excess voltage unless it was connected to ground. I've tested this circuit with and without the resistor and it performs identically with no difference on the O-scope waveform.

2) The original doesn't have a free wheel diode on the output.

3) Tried various caps from 510uf to 2000uf on the final output and it doesn't seem to make any difference.

4) I've found there needs to be a significant value resistor and diode between the pot and momentum cap, otherwise the voltage builds too quickly during accel, and when the throttle is turned to zero the cap will back drain through the pot.

5) There is a long delay from turning the throttle on til there is any sort of voltage on the O-scope. Because of so many transistors inline each with their own voltage drop the cap needs 3-4 volts of charge before the final transistor shows any output at all. Looking to add resistance "below" the pot to maintain a base charge in the cap at all times.

6) The push button throttle version doesn't appear to have any memory, so one would have to periodically press accel to maintain the charge in the momentum cap to maintain speed.

 

[diesel]

1) The zener/resistor circuit is just an output current limiter (to prevent overload if the output is shorted).
2) A freewheel diode is a VERY GOOD IDEA. Even if it will work without it.
3) Yes, it wouldn't make a huge difference.
4) Yes, the pot needs some protection from capacitor current. Less so, now that the cap is 1/10 the original value. I have a resistor R1 in place, the value can be changed experimentally. I didn't think about adding a diode, can you explain further?
5) Hmmm need to think about this. I won't have the assembled board for a few days, so I'll have a look when I do.
6) This might be better now that the control circuit has a FET

 

[diesel]

 

[diesel]

Still waiting for parts... my Digikey order seems to have hit some kind of international shipping snag. I can call tomorrow if it doesn't change. Generally shipping is 24 hours though. I already have the relay and socket installed on the PCB. I did wonder about the "dead zone" but it occurred to me that it would actually be helpful to have a small one. That is another thing I will have to experiment with when I get it working.

J2 is the control port, J3 is just auxiliary contacts for any direction indicators for example. J4 is to the track of course. J1 is for an AC adapter, about 10 - 16VAC should be about right.

 

[gasnbrass]

This is the breadboarded circuit I'm using so far:

I'm using just one 3904 as I read a report from someone who built several that the darlington paired 3904's would burn out. I only had 56K resistors on hand for high resistance areas, will need to get a variety of values to fine tune things, but the accel and braking works pretty good as is. The diode off the pot is to "trap" the momentum voltage to simulate how a locomotive will gain inertia and continue to role a long ways even with the throttle at idle. I am using a freewheel diode, I had an inline diode to prevent back voltage that could damage the main transistor but it seemed to dramatically reduce the output voltage.

This is an updated plan I need to get parts for. The variable momentum pot will affect both accel and decel, the braking resistors and momentum value would need some experimenting to pick a reasonable values. Could even add a third or fourth brake rate button. I'd swap the 3904 for a more common higher rated 2222. Plan is to have the throttle, direction, variable inertia, brakes and e stop in a wired handheld. I may add a bypass switch around the speed pot to cut the pot resistance in half to rapidly charge the cap and release as soon as the loco starts moving, or resistance below the pot to keep a constant charge. Currently I need to turn the pot high to speed up the charge rate and then turn it down when it starts to move, but a lesser value resistor between the pot and cap just lets it speed up too quickly when any speed change is made once it's moving. The direction control would be a relay affair, just didn't bother to model it at the moment.

 

[diesel]

I think I know why the driver transistors fail. It's a flaw in the output current limiting circuit. The intent is to shunt drive current away from the base of the TIP120 if an overcurrent produces enough voltage drop across the 1 ohm resistor. That works, but has a side effect. With the throttle cranked, and a short on the output, Q2 (in the original circuit) is not prevented from passing a significant amount of current into the Zener. That could be bad for the transistor or the Zener. I'm thinking about how to reconfigure it to avoid that.

I'm not sure about numbers yet, but your removal of one of the transistors lowers the gain, and therefore the short circuit current in your equivalent, the 2N2222. That would help, but maybe there is another way.

 

[diesel]

I believe, I goofed and the Zener is connected in the wrong place in my design. It should be on the base of Q1. I can patch that on the PCB for now. It forces a different version of what I said about the transistor failure. Maybe the resistor value could be higher so the transistor doesn't pass as much current in output short condition. It should look like this:

 

[gasnbrass]

I changed my schematic in Falstad to match the zener tapped after the 220R, but when I run the simulation using the dead short switch on the far right, the current through the load obviously goes to zero (left scope) but the current in the final transistor and the 1ohm (right scope) jumps way up, and it shows no current movement in the zener, so I'm still not quite understanding how it's supposed to limit the current into tip120. Also there was a note by the designer that in the production throttle they used a 0.5ohm instead, and various zener values for different current ratings, not sure if that makes a difference. I would think a simple PTC resetting fuse at the rectifier would also work, as I've seen in MRC throttles.

As far as the original Q1Q2 2n3904 burning out, they can only handle 200ma, and in a darlington pair that might exceed their rating in Q2 if too much current is fed into the base of Q1. The builder did say they ran fine for several years before the 3904's all slowly failed. The 2n2222 can handle 600ma so I would think a single 2222, despite the loss of gain, would be a more durable setup, or perhaps substituting a darlington pair of 2222 and see if it lives. I have tried both the single and darlington 3904 and it seems to drive the same, but I guess ultimately it would be a reduction in total power output using a single bjt. Not deal breaker for me, since I only run very efficient Kato's one at a time. But I'm an electrical amateur so who knows if I'm even thinking about this correctly.

 

[diesel]

More on the darlingtons later... but the output limit circuit can be guessed at. With the output fully shorted, on a half wave the voltage increases until the transistor chain turns on. There are 3 diode drops in the transistor chain, so approx 0.7*3 = 2.1V. The Zener conducts at 4.3V so it will shut down current to the output as soon as the difference (4.3 - 2.1) = 2.2V appears across the resistor. It does that by "stealing" the TIP120's base current. By Ohm's law, it would then set a current limit at 2.2A. Mainly, the current limit is to protect the 3055 from thermal runaway and destruction. One advantage over a PTC (comparing head to head functionality) is that it's instantly self resetting, it will return to normal after the short is removed. So an intermittent short like a points bridging, won't take down the power supply. You can change the 2.2A limit to anything you like by changing the Zener voltage or the 1 ohm resistor (but 1 ohms gives a good range, shouldn't be changed really...).

The reason you don't see much current in the Zener, I'm not sure, but you should see the current from the 220 ohm resistor. If you see absolutely no current, it's possible the Zener voltage on the virtual component is set way too high and it's just not conducting (at the same time, the current limiting would not be working at all).

This current limiting circuit was very common in hifi and other sound amplifiers in the 20th century. Sometimes they used a chain of diodes instead of a Zener, to save money.

The PTC itself isn't a bad idea, however since the circuit already has current limiting, some thought should go into choosing the right trip values for the PTC. In other words, pairing it with the output behaviour...

 

[gasnbrass]

So I think I'm starting to understand how that zener limiter bypass works. You mentioned a sting of diodes so I just chained a bunch together with a switch on the end. I simplified the right side to a freewheel diode, a 10ohm resistor and 10mh inductor to simulate a motor, with a 0.1 ohm "short" on a switch. The scope is measuring amps through the 1ohm sense resistor. With the short turned off, it made no difference in the amp draw if the diode string was connected or disconnected (small amp pulses on left). With the short turned on and the diode string connected, it limited the amp draw to the middle pulses. With the short on and the diodes disconnected, the amp draw shot up to the pulses on the right. Thanks for explaining how it works.

I also looked into how PTC's work and they apparently need quite alot of current to "disconnect" quickly, and will weaken with repeated activation, so probably not the best device to solely rely on for intermittent track shorts. If the zener bypass limits current to say 2A then a 3A PTC could be installed as a backup in case something internal in the throttle shorted out or the zener fails.

 

[diesel]

Okay, got it all together but no time to test it yet

 

[Kusojiji]

Nicely done!

 

[diesel]

I ran a test with an N scale loco. Unfortunately, my AC adapter is only 8VAC which is not enough, full throttle just starts a slow move. Also it was not enough voltage to operate the relay. I did notice the large dead band at the start, but it was expected. I'll try to find a proper AC source and scope it for the next test.

 

[shortwrench]

I ran a test with an N scale loco. Unfortunately, my AC adapter is only 8VAC which is not enough, full throttle just starts a slow move. Also it was not enough voltage to operate the relay. I did notice the large dead band at the start, but it was expected. I'll try to find a proper AC source and scope it for the next test.

A typical doorbell transformer is 16 volts AC. Easily found at your local hardware store,
reasonably priced.

 

[gasnbrass]

I power my test throttles with an MRC Twin Power Box transformer, 12vac and 18vac outputs, 65VA capacity and it has an internal circuit breaker.

That PCB setup looks so cleanly setup.

 

[roadbear]

Looks really good so far very professional
I've used this one in the past very similar I think

https://www.classixaudio.com/821.HTM

 

[diesel]

PWM throttles were also around when the CoolerCrawler was introduced. They're better than straight DC, but they lack the "capacitor magic" of the CC. The capacitor provides a kind of compensation for varying track/wheel resistance. On a (120Hz) half cycle where the resistance was greater, due to a dirt spot on the track or a motor commutator sag, the capacitor gains a little extra charge. On the next cycle, it will dump that extra power into the load. So it's a kind of adaptive feedback. It's a precursor to the modern motor feedback monitoring that you see in DCC decoders. It's conceivable that the CC capacitor effect could be combined with a PWM throttle, but it hasn't been yet.

 

[gasnbrass]

It would be interesting if the reactive capacitor could be combined with the waveform of the ITTC 3000 throttle. Square PWM like the Varipulse throttle create alot of noise in the motor. The sine wave pulse of the CC creates some noise, not as bad as PWM, but still there. The ITTC uses a fixed duty cycle pulse width that has low slew rates and curved edges, basically it has a slope that ramps up, curves into a flat voltage for about 50% duty cycle, then slopes down to zero. It controls loco speed by raising or lowering the flat voltage level instead of the length of the duty cycle. It's about the quietest waveform I've tried besides flat DC. I have no idea how it's produced in the throttle, I haven't tried to trace the board yet. If a reactive capacitor could be adapted it would be awesome, but I suspect that it would require an actual feedback circuit and microprocessor. The beauty of the CC is it's simplicity.

Varipulse

Cooler Crawler

ITTC Innovative Train Technology Control