Help building this 100 watt ss amp needed

[smackoj]

I bought a PCB for a diy 100 watt amp off a guy on ebay. I had some questions about mounting the transistors and I posted the schemo on Hoffmanamps.com forum which I follow and contribute to. I got a response from PRR who is one of the moderators saying "that amp is a cummy design that should be put in the trash"  I'm paraphrasing here. I would still like to have a go at it since I already bought power chips and driver chips and lots of 2 watt Rs. I'm putting the link on here to the discussion over at Hoffman's and would appreciate any other, 2nd or helpful opinions about what I would need to change or add to operate this amp safely?  thanks friends  :tu:

http://el34world.com/Forum/index.php?topic=17002.0

Sorry about the very poor quality of the schemo.

[teemuk]

- For 100W output, at least another pair of parallel output transistors
- Reliable and stabile therma-compensating bias circuitry. This will be difficult to realize properly because the output stage of that circuit has voltage gain and amplifies it's thermal instability.
- Emitter resistors to output transistors
- RC zobel to compensate inductance of the loudspeaker at high frequencies
- Some form of short circuit protection
- probably some close vicinity decoupling caps and high frequency feedback caps, depending on overall layout and stability of the circuit in such

...for starters. Though I have to say that I generally do agree with PRR, and you'd be better off building a good design from the start instead of trying to CPR a weak one that is obviously borderlining reliability and stability.

[bluesky6]

PRR sounds rather arrogant/ignorant with the following: "The days of making an audio POWER amplifier under 70 Watts out of separate parts are long-gone, and gladly. Make a necklace out of that PCB and buy a LM3886 kit off an asian eBay supplier. IMHO."

There may be a thriving industry around chipamps, gainclones and the like, but there are hobbyists out there interested in building audio amps with discrete components.

Google for John Linsley Hood or Class A.

There is a Thai manufacturer of electronics kits called FutureKit. They have a 100W kit (FK666). You can find a reseller on Ebay. Schematic found here: http://www.funnykit.co.kr/bemarket/shin/menual/fk666.pdf

I don't have that kit but have their 50W one built around a 2N3055/2N2955 pair. Just because when I was a wee lad, I used to dream about 2N3055-based power amps... and that's also for that reason I'm keeping the kit a kit ...

[Roly]

I am planning to use these power chips: BD243C and BD244C. They are not very big but, according to the data sht., will dissipate 65 watts each. The BOM calls for a 2n3055 and MJ2955

Let me tell you from practical experience that the best you are going to get out of a 2N3055/2N2955 pair is 50-60 watts into 8 ohms.  You can't take them higher because of voltage limits, and they don't like driving 4 ohm loads.

This is what a real 100 watt guitar amp looks like;   :dbtu:

http://sound.westhost.com/project27.htm

I can also see several other shortcomings or potential problems, e.g. the op-amp specified is a 741    which was (ahem) "quaint" thirty years ago - maybe more ... I forget.

The web abounds with circuits that have never actually been built, or somebody is telling fibs about what they can do, and this looks like another such.   :trouble

The ratings of a 2N3055 in a TO3 case are 15 amps and 115 watts.  The BD243 is only 6 amps and 65 watts, so they fall a long way short of transistors that won't do it either.  In a realistic circuit these maximum ratings have to be cut in half - or even less.

Strictly the 115W 2N3055 is a diamond or kite TO-3 case.  The TO-220 "2N3055" is actually the MJ3055 and it is de-rated to 90 watts in that package.

I have to agree with @PRR, except that building out of discrete components isn't entirely a passed art.


Some ground truthing;

Vsup = +/-34V

Assuming no losses (and you are actually going to lose a volt or two Vsaturation in each transistor), but ideally;

34Vpk = 34/1.414 = 24Vrms

24Vrms in an 8 ohm load

P = E^2/R

(24*24)/8 = 72 watts at the onset of clipping.

So even downhill with a following wind "100 watts" is simply bulldust, not a hope.

I have a proper circuit using 2N3055's with +/-35V supplies in my Twin-50 that delivers just over 50 watts into 8 ohms at the onset of clipping.

Just out of curiosity I did an LTSpice sim on this and it clearly has some serious problems.  I think teemuk can put the defib down - this one is DOA.

[smackoj]

Wow! great input friends. I don't really care about the pcbs I bought from the 'masked man' in Thailand cuz those can be plunked in the pot. I do, however wish to build a nice power amp and I have lots of parts left over. I couldn't find the 2955 and 3055 chips with the newer style package so I bought these chips:  BD249C and BD250C. I bought 2 pairs so I have 2 of each. I bought the listed BD139 and BD140 driver chips and I have quite a good selection of single and dual op amps. Would anyone like to offer a suggestion  on what design I could/should build that would work good with these particular transistors?  Note: I had seriously looked at the ESP 100 watt board before I bought these boards. I guess with my Gear Acquisition Syndrome habit of late, I just wanted a simpler, less parts count idea to start. Live and learn as the old farts like to say.

For those readers that may want to scope the original schemo, here is a better viewable copy that I added some highlighted parts to 'make it easier' to see/read.

thanks again friends, lots of good info. I appreciate it.   

[J M Fahey]

Here, I cleaned it for you.
Or rather for me, to be able to at least *see*  the circuit

*IF* you already have *everything* , including transformer, cabinet, etc. just build it.
It will probably work .... within its limitations, meaning some 50W RMS into 8 ohms (forget 4 ohms).
At least as solder practice.

Now if you only have the PCB and whatever goes into it, but no transformer / cabinet / chassis / heatsinks, etc. the most expensive parts by far, then don´t, just set your sights on a better project.

Following the old: " do not waste good money after bad"

[smackoj]

thanks for the 'clean up' job. I think I'll just try to find a better design. I don't really want a 50 watt amp and I get lots of practice soldering with my pedal projects. I keep lots of spare 'recycled' transformers and other parts around so I won't be buying any of those pieces NEW.

Have any forum contributors used the BD249 and 250 chips in a good design?

thanks amigos    :tu:

[J M Fahey]

Well, you can use them here:
http://sound.westhost.com/project3a.htm
You´ll be safe with +/-35V and a single pair or up to +/-42V and a double pair, all well heatsinked, of course.
http://sound.westhost.com/project27.htm

There are certainly other 60 to 100W amps using complementary pairs, main problem is getting the PCB or at least the design to print and etch them.

If you find an interesting one, post it here.

PD: and as said before, if you have everything to build this one, just do it
Typical use for such unprotected amps is to build them inside a speaker cabinet, to solder it straight to the speaker terminals, no jacks/plugs involved.
Or use those push-in red/black connectors popular in home audio, so it´s easier to avoid shorts.

If you do so, post a few pictures of what you have , specially PCB, chassis  and heatsinks, so we suggest safes way to build it, and probably a couple small tweakings (such as a Zobel network, diode thermal coupling for stability, etc.)

You might end up having a powered cabinet, great to test your pedal projects, some guitar/bass/mic preamps , listen to some music at the Lab and even use it live (with a proper preamp or driven from a pedalboard).

And absolute worst case, if someday you blow the output transistors, you junk them and the PCB and mount a TDA7294 kit using heatsinks, PSU and chassis.

You are not wasting your time .

[Roly]

I'll repeat my point; the BD249C and BD250C have about half the power rating of 3055/2955's, and if the best you will get out of them is 50 watts into 8 ohms, then the BD249C and BD250C are going to give you at best about 25 watts out per pair.

It's not often that I disagree with Mr Fahey, but in this case I think that building the original "100W" circuit is a waste of time.  If nothing else it starts out with a blatant lie, and I don't think that's a hopeful sign at all.  I'm not sure how far I can trust the LTSpice sim, but it suggests that the output pair have a real problem at crossover time and produce large voltage and current spikes.

I haven't investigated the cause but I assume that this will be related to the slowness of the '3055 family and the lack of Base-Emitter resistors causing both output transistors to be conducting at the same time due to "hole storage" in their Base-Emitter diodes.

I'm sure I have seen this circuit somewhere before, possibly as a reader submission to Wireless World magazine, but can't place it, but I do remember dismissing it as a horror at the time.

Lacking meaningful temperature compensation I expect that it will eat output transistors for breakfast.  There are many other circuits that IMO have a much better prospect.

Following the old: " do not waste good money after bad"

a.k.a. The Sunk Cost Fallacy - "The project has gone pear-shaped, but we've spent so much we can't stop now".  A classic example is the new Miki public transport ticketing system here in Victoria that has already cost well over a billion dollars and still isn't working properly, but nobody has had the guts to face the horrible reality and pull the plug, so millions more go down the drain chasing a reduced spec they still can't meet.

[smackoj]

sorry about having to hear it 2 or 3 times. I thought that 2 pairs = apprx 100 watts. I get it now; equals apprx 50 watts.

I'm ok with not building this 'too far gone' design. No sweat  :tu:

[bluesky6]

I didn't realize it but it looks like same Ebay seller who sold smackoj the "100W amp" PCB also sells the FutureKit 100W amp.

That same op-amp driving power transistor design is used in a kit by a US-based seller (xkitz). This kit is based around the TIP41/TIP42 pair and is advertised at "30W".

Although I may not agree with PRR's views on discrete audio amps, I would agree that the best way to achieve 100W in a home-built would be something using a couple of chipamps in parallel mode or better still, one of those Class D amplifiers that are also advertised on Ebay. The Vox VT80+ uses one of those, for example. I am also pretty sure that the ZT 200W shoebox amp uses the same approach.

For laughs, the attached picture is that of a transistor-based 100W stereo kit on my Ebay watchlist.

Notice the size of the heatsinks...

[smackoj]

<notice the size of the heat sinks>

They appear on the 'too small' size to me. Is that what you mean by "look at the size" ?

[Roly]

At 65 watts and 6 amps the TIP41/2 pair are similar to the BD249C and BD250C - about half the ratings of 3055/2955's


Those "mini TO-3" heatsinks have a thermal resistance of around 5°C per Watt where a genuine 100 watt amp would require a heatsink with a resistance of no more than about 0.5°C/W, and the lower the better.  These wouldn't even cut it in a domestic stereo which only peaks to 100 watts occasionally, and certainly not in a guitar amp which is much more arduous duty.

This is rather more what I'd expect;

0.39°C/W  It's about a foot long (300mm), anodised black.

Details;
http://www.altronics.com.au/p/h0542-300-x-75-x-46mm-flanged-heatsink/

The power amps that we deal with are inherently about 50% efficient, so if you're getting 100 watts into the speaker there is another 100 watts of heat to get rid of from the output devices.

100 watts out is 50 watts in each transistor (assuming only a pair), and mini TO-3 heatsinks have a temperature rise of 5°C/W, so 50 * 5 = 250°C.  The typical maximum chip temperature is 150°C and accounting for the thermal resistance between the actual chip dice and the heatsink means the heatsink has to be somewhat cooler.  Then you have to remember that this is for free air at 25°C ambient, and that every degree above that will add to the final chip temperature (which is already well into meltdown).  If you only do outdoor gigs in Alaska you might be fine, but even here in southern Australia, on stage, under lights, on a hot summers night we might be talking 50°C ambient and the chips would be melting.

One of the most common mistakes that amp builders make is overlooking this, underestimating, and only providing the pathetic heatsinks shown on the kit.  Cheap, light for shipping, and useless.

When it comes to stage amps there is no such thing as "too much" heatsink.

30 watt amp?  Yeah, possibly, perhaps with a fan maybe, but I'm still not thrilled.  100 watts?  Never - or not for very long, anyway.

[phatt]

Worth a read,, though it might take a while.
http://sound.westhost.com/heatsinks.htm#s2
If you buy and Build ESP products you at least get backup.
A nice thought when starting out on something you are not quite sure will work. :dbtu:
Phil.

[smackoj]

Nice stuff about thermal transfer from the ESP site. Thanks

:tu: