Vbe Multiplier for LM4702 Output Stage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Short question:
How do I go about choosing a transistor to serve as the Vbe multiplier to avoid thermal runaway issues? What specs do I want, and how do I go about matching them to the darlingtons I have chosen?

Long Question:
This is my first post here, but I've been reading for a while, trying to absorb as much information as I can.

The LM4702 will be my first project beyond a chipamp, so I'm hoping I can learn a bit about choosing transistors. I would like to build a (relatively) simple amp using the typical circuit schematic in the LM4702 datasheet. It's quite possible that this project is too much for me to handle, but I'm hoping that by narrowing my scope to just the output stage, I'll be able to do it. If this question indicates that I'm in over my head, please let me know.

I recognize that most people have chosen to ignore the recommendation of the darlington output stage, but I'm looking for a learning experience, and not necessarily the best amplifier. Anyway, I've run into this issue and I don't really know where to go to find a solution.

I understand the thermal runaway issue, and I follow National's recommendations of using a heatsink and emitter degeneration resistors (page 11), and I also understand how a Vbe multiplier, or Qmult as they refer to it, can help solve the problem. How do I choose a BJT to act as my Qmult?

Also, if anybody knows of any websites or books where I could learn more about this, I'd love to check them out too.

Thanks in advance,
Robert
 
Robert -- I was able to purchase the Sanken power darlingtons from Profusion in the UK -- I received them in less than a week -- here is link:
http://www.profusionplc.com/pro/gex/prodGen.html?prdtyp=power darlington#

beefy little guys with tempco built in.

this is a schematic -- showing one channel -- using the LM4702a and SAP16's:

An externally hosted image should be here but it was not working when we last tested it.


fwiw -- I have 2 versions of the PCB on the way -- one sites the transistors along one vertical plane, the other on two parallel vertical planes.
 
The Vbe multiplier transistor is not critical. Almost any general purpose NPN small signal transistor can be used! It must be mounted on the heatsink near the power output transistors to be effective. Its Vbe needs to be able to follow the thermal variation of the Vbe in the output transistors.

BTW, National Semiconductor have just published a new Application Note #1490 "LM4702 Power Amplifier" in May 2006. The 10 page PDF is available at
http://www.national.com/an/AN/AN-1490.pdf
Good reading!

Glenn
Melbourne, Australia
 
AN1490

Hi,
I notice the following about the paper:
1. Cin is a Panasonic FC, not a non-polar cap as drawn on the schematic.
2. MN2488 and MP1620 are long obsolete. So are 2SD2488 and 2SB1620 (There must be a dusty warehouse somewhere in NS).
3. No V/A rating for the transformers.
I would also say that there should be overload protection and relay, but that is personal I guess.
I would trade my tin boxes for the case any day though.
Cheers,
KK
 
Since we are talking about it here, the big things I noticed from the paper were the lack of a heat sink on the LM4702, the choice of feedback resistor value, and lack of coupling caps.

1) I haven't noticed any heating while testing my LM4702's but I have been using decen sized PC heat sink. Has anybody tried running the chip without cooling? How does it work? I would guess based on quick, back of the envelope math that it would probably be OK, but I don't really want to be the one to fry the first chip. Any takers? :D

2) The feedback resistors are a little higher values than I have been using. I like the idea of using a 2.7K resistor in the input path, but the 68K resistor seems a little high. Has anybody directly compared different values yet? I personally haven't gotten around to it since I have been focusing on the output stage.

Also, the combination of these resistors sets a gain of 26.2, which is right at the limit of what National recommends. I have been keeping my combinations around 30. Any comments about what values have worked best for everyone else?

3) I LOVE the fact that they got rid of the coupling caps!!! I have been trying a bunch of combinations and figured it was just my lack of trying enough to make me happy. I am amazed I didn't just chuck these things a long time ago.

I also like the 200uf Ci, and have played around with some values, but haven't really gotten that high since I have been trying to keep Cin in consideration too.

Finally, I wish this note had come out before I built my boards, but fortunately they comply with most of what was said in the app note! For the couple of things I would consider doing differently based on this paper, I can easily run a couple of jumper wires to get back on track.
 
dfdye said:
Since we are talking about it here, the big things I noticed
Finally, I wish this note had come out before I built my boards, but fortunately they comply with most of what was said in the app note! For the couple of things I would consider doing differently based on this paper, I can easily run a couple of jumper wires to get back on track.

before i did my LM4780 Bridge and Parallel boards I went through several iterations of prototypes -- exposing and etching your own boards is pretty laborious -- but it's the only way to make sure that some odd coupling leading to oscillation isn't going on, that there is no modulation from the power supply rails etc. then once you get your boards back you wire one up, hold your breath and run the analysis to make sure that it behaves as the prototypes did.

wrt the PdMax for the LM4702 -- figure that you have to dissipate about 4 watts --
 
Well, jackinnj, the boards I made are up and running on my bench just fine, but I included several things and positioned several components differently than I would have. Sorry if there was confusion that this somehow changes how they perform.

Also, the National guys even say you don't need a sink for the IC, so I am going to take their word for it, but verify! :D

Tellingly, though, I have never had the sink I had attached to the IC ever get anywhere even close to warm, so I am quite ready to belive that passive cooling without a sink will be sufficient.
 
they specify Tjmax of 150 degrees -- I would use a somewhat lower value -- at any rate the heat sink is not very large -- a C/W rating of 30 would be fine -- so an off the shelf TO-220 variety could probably be pressed into service.

the real test of a board is a THD + N% test at 1 watt output -- from 10Hz to 50kHz.
 
jackinnj said:
at any rate the heat sink is not very large -- a C/W rating of 30 would be fine -- so an off the shelf TO-220 variety could probably be pressed into service.
Absolutely. I think we are on the sampe page with this one. However, a standard TO-220 sink is too narrow to fit the chip, if memory serves. . . . I'll have to check it out, but it would make things very easy if it fit.

David
 
dfdye said:
Absolutely. I think we are on the sampe page with this one. However, a standard TO-220 sink is too narrow to fit the chip, if memory serves. . . . I'll have to check it out, but it would make things very easy if it fit.

David

look in the DK catalog -- there are TO-220 device heat sinks for the wide devices -- what is important is that the metal on the back of the LM3702 be in complete contact with the heat sink.
 
jackinnj said:
. . . what is important is that the metal on the back of the LM3702 be in complete contact with the heat sink.
I am more than aware of that aspect of heat sinking, but thanks for the tip. If something is really critical, I won't hesitate to resurface a heatsink on my mill and lap the thing untill it shines. I know the surface of the chip is the limiting factor at that point, but I guess I am not quite hard core enough to lap the IC's too.

Still, considering the heat of the LM4702, I don't think any of this is necessiary. Any simple heatsink should suffice, even if you just scotch tape it in place! :D (Time for testing!!!)

David
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.