PDA

View Full Version : Source for AB power transistor



Jim Caudill
09-12-2007, 11:37 PM
I posted this same request over at the PM forum, so forgive me if I seem a little desperate. I have this AB power transistor number, but have been unable to locate a cross reference or source to purchase same. The number that I believe to be AB's is 245D00031E. Since the device came from A CMC Servomate SM1, I have the schematic and it is listed as component Q7 bearing their part number 29782LG. I know there are some real electronic guru's here so, I may have a better chance at success with you guys.
http://i72.photobucket.com/albums/i183/Ferrofab/MVC-019S-1.jpg

JoeFin
09-12-2007, 11:48 PM
Custom Part Number

AB doesn't make power trans, I can just about guarantee you that. What they do do is to buy a large lot and have their own custom number and logo printed on them.

Keep checking for a cross ref. trans that well fill the bill or just plain break down and call AB and talk to their service techs. If by any chance they have discontinued the use of that particular part they will be more then happy to give you a replacement (stock) part number

But....

In the past I had this problem with Motorola SCRs which were very rugged and well made. The problem was there were no more of these Motorola SCRs available. The replacement held up electrically, performed beautifully, BUT did not have the same heat dissipation characteristics of the original.

Simple fix, I had to ad a fan to the power amp is was in and the thing worked fine

J Tiers
09-13-2007, 12:13 AM
That is absolutely an Allen-Bradley in-house part number, and bears no relation to any other numbers, as far as I can determine.

Failing getting the part from them, the only recourse is to reverse-engineer it, looking at the schematic and usage of the part to derive the requirements for a suitable substitute.

In these days, when bipolar transistors are getting rare, especially in the TO-3 case, that might be challenging.

A glimpse of the schematic might prompt some more questions and get to a sub......

What kind of circuit is it from? Something that is potentially a switching type drive or power supply? Or a "linear" type circuit? The latter will be easier to cross, as switching bipolar transistors 'went away" longer ago than linear.

Jim Caudill
09-13-2007, 01:22 AM
So far everything I have tried has come up empty, including Allen-Bradley (now part of Rockwell Automation) parts department. AB has created a "case number" and will have technical support research the part number and call me back if they can find out anything about it.

dp
09-13-2007, 01:58 AM
Post the part of the circuit it is found in. It may be possible to cross-reference it. If it's an old germanium transistor you may be SOL but if it's simply a common NPN or PNP switching power transistor and the voltages are known there may be many substitutes including the ubiquitous 2N3055. If it is in a balanced analog circuit then it may still be substitutable but should be swapped out in pairs.

Just be warned - without seeing the circuit it is in it is not possible to know with any certainty that it is even a transistor. It could easily be an SCR or even a diode in sheep's clothing.

Jim Caudill
09-13-2007, 02:37 AM
Here you go. Note that the "transistors" in the top half are numbered from left to right as Q5,Q6,Q7,Q8,&Q9. The bad transistor is Q7.
http://i72.photobucket.com/albums/i183/Ferrofab/Sevomate1.jpg

dp
09-13-2007, 02:45 AM
That is a common NPN switching transistor. The only voltages shown are +12 so it may simplify things. I'll see if I can wrangle up an available replacement and perhaps others can be looking too.

Edit: As it happens, the 2N3055 transistor just might be a good substitute. It is a hardy switching/Audio transistor and has a very long pedigree. Q7 in your circuit is running complementary to Q12 so I'd be tempted to change them both at the same time. A fundamental nature of switching transistors is that they saturate quickly so have few power dissipation issues, and that they have low vce levels which is the voltage drop across the emitter/collector junction. Anyway, it's available, or was, at radio shack and worth a try.

Evan
09-13-2007, 05:35 AM
If it's NPN in a TO3 case a 2N3055 will often replace it. I have a handful of them in my parts stock. They should still be available.

J Tiers
09-13-2007, 10:11 AM
If you look closely at the circuit, it is rather interesting. Things may not all be as they seem......
It looks as if there might be a need to see what is off to the right before making a final decision on what the voltages really are. The 12 V is not the supply for the part in question. A 2N3055 and maybe even a 2N3773 would likely not be a good choice.

The value of R102 is a hint, along with the value of C39.

Jim Caudill
09-13-2007, 12:49 PM
Here is what is to the right of the portion displayed prior. It is a 90vdc servoamplifier capable of delivering 12amps continuous and 20 amps peak (if any of this helps)
http://i72.photobucket.com/albums/i183/Ferrofab/Servomate2.jpg

Wirecutter
09-13-2007, 02:26 PM
Another interesting thing to note: Other transistors in the circuit, such as Q6, Q8, and Q9, have more "standard" 2N-style numbers. This suggests that Q5, Q7, Q10, and Q12 are indeed special in some way. Just a thought.

-Mark

Jim Caudill
09-13-2007, 02:36 PM
I spent some time with CMC tech support today and learned the component is no longer available. He indicated that when last sourced, it was a "darlington" component (is this a mfg?) and that it is a "fast-switching, cascoding" type of transistor, NPN in a T03 case. He felt the 29782LG number would be the best one to try and match, although he did give me the last CMC number (X18-19246). Using the LG number sort of makes sense, since the adjacent transistors have numbers like "2N5322". He also stated it was being used in a "push-pull" configuration and it does look like the 2 transistors together are operating between the 100vdc circuit. I'm concerned about just how difficult this may turn out to be. He made the comment something like "well, you said yourself, this is over 20 years old" like I'm supposed to scrap a nice running Hurco over a single transistor? I've got some other leads, regarding old servo amps, that I'm working on.

dp
09-13-2007, 02:37 PM
Another interesting thing to note: Other transistors in the circuit, such as Q6, Q8, and Q9, have more "standard" 2N-style numbers. This suggests that Q5, Q7, Q10, and Q12 are indeed special in some way. Just a thought.

-Mark

It could be that what makes them special is that they are matched sets. I used to get matched transistors from TRW for hi-power RF amplifiers I manufactured. That matching resulted in new part identification. They were batch matched meaning I could grab any two transistors and be assured they were sufficiently identical in the important areas as to be interchangeable.

dp
09-13-2007, 02:45 PM
I spent some time with CMC tech support today and learned the component is no longer available. He indicated that when last sourced, it was a "darlington" component

Interesting - it isn't drawn as a Darlington but that is often the case. It does make it difficult to reverse engineer old electronic equipment when the components are not off the shelf. Darlington transistors are actually two transistors in a single chip. They share the collector terminal but the emitter of the input transistor is wired directly to the base of the ouput transistor in a common emitter configuration. It reduces the drive current needed to saturate the output transistor. It remains a three-wire device. In switching circuits these are often matched sets.

Wirecutter
09-13-2007, 02:51 PM
Ooooh, I may be able to help here. My stepdad once had a company that used matched transistors, and I sold his transistor curve tracer for him on Fleabay.

One thing stepdad gave me was a couple of ziplock bags full of TO3 transistors that had been characterized for hfe and breakdown voltage. These were used in parallel (hence the matching) in a 160v, many-hundreds-of-amps power supply. They may just be rated beefy enough for your use. If the stars align properly, it would be no problem for me to pick 4 (matched spec units) out of the bag and perhaps trade you something for them.

I'll check the specs on them when I get home.

Jim Caudill
09-13-2007, 03:26 PM
Latest information I got was from Buddy at AMTS. He stated that I could substitute Motorola part number MJ10021, but I need to replace all 4 at the same time and that they should be from the same batch (date code). Naturally, he is out of them and would like for me to tell him if I can locate a bunch at a decent (less than $10) price. Any leads here? In case somebody does a google search I'm posting all the numbers here together. Allen Bradley 245D00031E, AB 245D00031E, 28782LG, CMC X18-19246, Motorola MJ10021 must be matched set of 4.

J Tiers
09-13-2007, 11:59 PM
Yep, the MJ10021 ain't just "any" darlington compound transistor.

The top of the data sheet says it is a "switchmode series" 60 ampere 250V power darlington transistor with integral base-emitter speedup diode. It is a fast device, characterized for a bunch of parameters which may or may not be needed in this application.

It does not look like some of the special features of the MJ10021are being used in this particular circuit, which appears to be a "full bridge motor driver". But without knowing a bit more about the unit it's hard to tell.

Apparently the 4 of these parts are pulse-driven by the 4 transformer coupled base drive circuits. But evidently this is not at a very fast rate, due to the diodes in the drive circuit.

The matching may not be for the usual characteristics. it may be for on and off times, to avoid "crowbarring" by having one still partly on when the one below it turns on.

Date code is better than nothing, but matching depends as much or more on die position on the wafer as it does on date code alone. Been there, had the trouble.

In any case, the part they mention makes sense, and you have a direction.

Jim Caudill
09-14-2007, 12:27 AM
I have order 8 of the pieces manufactured by ON Semi that should be here next week. I also have ordered 20 supposedly true Motorola units manufactured in Mexico (don't know how long ago that may have been, since Motorola has apparently gotten out of that line). The rub is, I ordered the 20 pieces from an outfit in Hong Kong, and I have been advised not to trust any Asian components as they print whatever they want on any similar component.

1st question, any idea how true the rumor is about getting components from Asia?

2nd: Is there a way to match the transistors for performance? How would you go about doing this?

Thanks to all of you for your time, expertise, and willingness to help me get my mill back up and running.

dp
09-14-2007, 12:45 AM
2nd: Is there a way to match the transistors for performance? How would you go about doing this?



For switching the important stuff is Hfe, and Vce (sat). That is the gain of the transistor and is fairly easy to check with common tools for the purpose, and the collector/emitter voltage when the transistors are saturated, that is to say, fully turned on.

More difficult is the switching time - apply a pulse to the transistor's base and measure the time it takes the collector to reach a voltage level - often the saturation point. And inversely, the time it takes to turn off. A dual trace O'scope can help but it's really lab stuff and involves a well calibrated signal source. For HSM kinds of things it can be done well enough with bench test equipment.

Optics Curmudgeon
09-14-2007, 01:29 AM
The traditional tool for matching transistors is a curve tracer, an oscilloscope specifically built to plot semiconductor characteristics. Tektronix made a couple of models. Also, NTE, in New Jersey, has these parts, although you never know where they came from.

Joe

darryl
09-14-2007, 03:03 AM
Looking at the circuit, it appears to be a somewhat straightforward power amplifier, with three separate outputs. Q7 and Q12 are a pair in one section. The complex base drive circuit has been designed to allow for only one polarity of output transistor to be used, being the NPN, which is the more rugged of the two types, or at least was at the time. It also allows for some isolation from the drive control circuitry. I would expect that some feedback is used, which would help to make the output follow the control signals and self-correct for errors to some extent. I doubt that matching the output devices would be required- most important would be to select the right type, whether darlington or standard, the right polarity, which is NPN, then allow for voltage and current ratings- a 250 volt rating won't be overkill in this case, and neither would a 20 amp current rating. Judging from the 22 uf cap across the base/emitter, it would not have to be a high speed transistor. Probably any transistor with suitable ratings would be faster than that original anyway.

I think you'll be ok replacing the devices by pulling any two from the bag, if you get them all from the same batch at the warehouse. Testing and matching them won't be easy unless you're competent in electronics, and have the suitable equipment. Even many electronically inclined people won't have the proper equipment to do a test and match devices. I would suggest don't worry about it.

Check carefully how good the connections are to the transistor pins. Often a bad contact in a transistor socket will be the cause of the failure. A poor connection to the emitter can put a lot of heat where it can do damage, and a poor connection to the base can prevent the driving circuitry from controlling the transistor properly. If the connections are through a socket, and not soldered directly to the transistor pins, then definitely check it out. You may need to find out why that transistor blew, and correct that, otherwise it may blow again. If you replace parts and it works, that's nice, but it would be better if you can nail down the cause of the failure, or at least the probable cause.

One cause which is common is poor thermal contact between the transistor and the heatsink. I've seen cases where the heatsink compound is so dry that it's virtually useless, and often enough the screws that hold the transistor down have loosened over time.

One thing bothering me is that the transistors are not drawn as darlingtons, and being transformer driven as they are, the drive current for a standard transistor could easily be supplied by that circuit. I can't see the need for a darlington there, and it's possible that they aren't darlingtons. I hope you've been given the correct information.

Jim Caudill
09-14-2007, 04:30 AM
Yes, I'm confident that I have been given the correct information. This is supported by both the factory tech support and an independent Hurco specialist (AMTS- Accurate Machine Tool Specialist, Buddy Maughon). The transistors are soldered in place and failure was initiated by an intermittent contact in a 16 pin connector made by AMP. The pin and sleeve connection for one of the x-axis servo motor supply leads and become somewhat separated due to the pin pushing back in the plastic housing. Arcing was evident between the pin and sleeve to such an extent that the pin could no longer enter the sleeve. The servo wiring harness had been "duct taped" to hold the cord in a binding position where the pin could maintain contact. Not recognizing the "duct tape" repair for what it was doing, I removed the tape and cleaned all the residual glue and tape from the harness and where it was tied. Suddenly I was experiencing an intermittent servo, and all sorts of "motion error" codes. It was during the "wiggling" of the harness that I caused the failure of the servo amp. A new plastic CPC housing was ordered along with the pins, sleeve, and extraction tool. The connection issue has been resolved, now it's time to get the servo amp fixed.