Hybrid amplifier.. Please help..

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
hey guy's.. i have been experimenting some more.. just came back and fouind your replies..
I moved Q3 directly on top of one of the output transistors to see if that made a differance.. but it didn't.. it is likley to be just it warming up and me being overly paranoid.. the heatsink is cold to the touch.. i moved the 10uF cap onto the wires on Q3.. its about 1cm away from Q3 now.. it stabilised the bias a bit (not jumping up and down so much.. changed it for a 22uF and it is rock solid now.. But the bias does keep riseing.. i dont have room on the current heatsink for the drivers.. but i will be changeing the heatsink when i do the final build.. i guess i should just leave it powered up for a while and keep an eye on the bias.. i was starting to worry when it was hitting about 30mV..
I'll have a play and see what happens.. :)
Should i leave Q3 on top of the output transistor or mount it back in between the output transistors on the heatsink itself? bearing in mind the BD139 has metal on the back of it so i have to use thermal tape if its mounted to the sink... whereas if i mount it to the output device i can get away with just useing a small amount of paste...
Thanks,
OWen
 
Hi Owen,
The thermal runaway issue can be tricky to tackle. I am a little surprised that you should have any sustained difficulty with such large (0.33 ohm) emitter resistors. So I tried to look up your output device datasheets. I couldn't find 2SC2579 but I did find 2SA1104. Unfortunately, the datasheet was very brief an gave no info about the sensitivity of Vbe to Tj (junction temperature).

What it did say, however, is that the device is rated at 80W power dissipation and 8A max current. I would advise, if you are going to use this circuit to its full potential that you change these devices. 80W is not enough with 30V power rails. You really need transistors rated above 150W, preferably above 200W. It depends on the speakers and listening levels, but bipolars are unforgiving if their SOA (safe operating area) is exceeded and their SOA becomes very restricted as Tj increases and Vce increases.

I know it is a pest to change devices but I would recommend you use something like MJ15022/MJ15023 which are rugged and have a good frequency characteristic. They will not runaway thermally with 0.33 ohm emitter resistors.

The thing about putting the drivers BD139/140 on the same heatsink as the power transistors may not be so good in some circumstances. The drivers MUST be on heatsinks to keep their temperatures stable - the TO-126 case is not good on its own for shedding heat. These devices will see much less Ic variation than the output devices and so can be kept at a fairly stable temperature if they have their own independent heatsinks. If they are on the output device heatsink and this heatsink temperature changes significantly with output device temperature then the bias current will be doubly affected - both the drivers and output devices temperatures will change and both their Vbe will decrease. In any case the drivers must be on some heatsink and putting them on the output device heatsink is better than nothing. You can get fairly small finned heatsinks especially for TO-126/TO-220 cases and they are quite cheap.

Sounds like you are making good progress.
 
Administrator
Joined 2004
Paid Member
Hi Owen,
whereas if i mount it to the output device i can get away with just useing a small amount of paste...
Is the output a metal case or plastic (can't remember). At any rate, you should mount the bias transistor to the heatsink close to the outputs. In the final build, mount the drivers on the heatsink also. For now, try small heatsinks on the drivers. That might steady things a bit too.

At the higher power levels, all these things will become more important. Bias transistor, drivers, bootstrapping or CCS. Everything.

-Chris
 
The datasheet I found showed the 2AS1104 in a plastic package, looking a little like TO247 but not the same. So I concur with Chris that Q3 needs to be on metal (with insulator) not on the plastic package or it won't sense the silicon die temperature well enough. The plastic is not a great heat conductor.
 
i dont have problems with thermal runnaway...it was my multimeter acting up :smash: on the 200mV setting it isnt working right.. on 2000mV setting the bias is rock steady and never rises from what i set it too.. infact it does go down slightly after the amplifier warms up a little.. the output devices are plastic case with a metal back. they dont even get warm to the touch.. the amp they came out of believe it or not was running + / - 45V rails (just went and measured them) i have however got a spare set i managed to get on ebay aswell.. :)

This is all i could find on the 2SC2579:
Part Number = 2SC2579
Description = Si NPN Power Bipolar Junction Transistor
Manufacturer = Sanken Electric
V(BR)CEO (V) = 120
V(BR)CBO (V) = 160
I(C) Abs.(A) Collector Current = 8.0
Absolute Max. Power Diss. (W) = 80
I(CBO) Max. (A) = 100u
h(FE) Min. Static Current Gain = 30
@I(C) (A) (Test Condition) = 3.0
@V(CE) (V) (Test Condition) = 4.0
f(T) Min. (Hz) Transition Freq = 20M
Status = Discontinued
Package = TO-218var
Military = N

many thanks,
Owen
 
also i have had it at full volume now for about 10 -15 minutes and nont of the transistors are hot.. the drivers arent even warm.. but i will mount them on a heatsink for the final build, only the output transistors are slightly warm to the touch..
I will use different output transistors for the next version of this amp i build. Mabey even try and get a bit more power out of the next one ( even though i really dont need any more power) The amp is running at 23mV accross both resistors now and i think it is sounding better than ever!!
Owen
 
Administrator
Joined 2004
Paid Member
Hi Owen,
I see your specs. From my experience. Amps that used these either had supplies that caved under load or they blew up. I fixed TONS of amps that used these. I installed 2SC2581 and 2SA1106. It wasn't worth buying the cheaper ones.

The heatsinks in those amps were normally too small as well, they depended on protection circuits to cut the load out.

-Chris
 
yeh, the heatsink was deffinatley undersized (u shaped piece of bent aluminium) I'm useing a different heatsink at the moment.. and i will be useign a better heatsink again in the finished amp.. Would you suggest i have a look see if i can source some 2SC2581 and 2SA1106 to be on the safe side?
Owen
 
i can get this one

2SA1106
(SNK) PNP 40V 10A TRANSISTOR *324

and this one

2SC2581
(SNK) NPN 200V 10A TRANSISTOR *324

but they dont apear to be matched.. one being 40V and one being 200V.. i assume the 2SA1106 is the wrong one? :S they are fairly cheap aswell but i think i will just leave it at the moment.. if i have problems with these ones i will change them.. but to be honest i am thinking bout useing different output devices in the next version of this amp anyway so for the short time this one will be getting used daily i doubt i will have problems..
Owen
 
Administrator
Joined 2004
Paid Member
Hi Owen,
2SA1106 is 140V, 10A, 100W. 2SC2581 is the compliment with the same ratings except C-B at 200V, it's the C-E rating of 140V that's important. I've seen some spec books with incorrect data on these.

These days, I'd be using MJL1302 and MJL3281 or whatever they have in that bigger package. For bigger amps try MJ2119(3,4 or 5,6). I understand that you were using those to prove a concept.

-Chris
 
Hi chris,
You are precisley right, i was just useing what i had here to prove the circuit before i spend any money on it..
There are a few things i have to order before i can do much more with it so i will throw some different output transistors into the order and have a little play and see what works best for me.
At the moment i'm just happy that it is finally working and that it sounds half decent, It is afterall my first dive into the realms of DIY solid state Audio.
I spose next thing i have to do is sort out the muteing relay to go between the tube and SS stages..
Would it be terribly wrong if i mounted the drivers ont he heatsink aswell and built it P2P on tag board? i was looking at it earlier and there are so few conponents that it would be easier to do this way and cut down on solder joints. the tube stage will be wired pretty much entirley ont he back of the tube sockets as there really isnt anythign there that even warrants tag board. Power supplies Both have their own PCB's anyway and these are already built. Or would i be better off just etchign a couple of small PCB's?
I spose the proper way to do it would be to make a PCB with the tube stage, the muteing circuit and the Output stage conponents all on one board.. and the board really wouldnt be that large if i did it this way so mabey i will do this, I'll have a play in eagle and see what i can come up with..
Many thanks,
Owen
 
Hi Owens,

If you connect a relay between the tube section and the SS section you'll need a cap between the TS output and the relay. Without the cap when the relay kicks in, there could will be a voltage spike equal to the TS dc output voltage (assuming the voltage across the 2.2uF cap is zero) on the other end of the SS input 2.2uF caps at Q2 and Q4 that could damage the amp, and possibly your speakers if connected. I figured I'd bring it up just in case. It's why I suggested putting the relay at the output, instead of using an extra cap, plus the advantage of speaker protection. And as long as there is no output load the BJTs won't sweat it as the tube section stabilizes.

Cheers,

Al
 
Actually, no, you don't need an extra cap, but you do need a relay with two contacts.
This is what you do: you put one end of each contact ontot he bases of the driver transistors. The other end you connect to ground). Use the normally closed contacts, so that they open when the relay is energized and the amp un-muted.
When the relay is in mute position, the input of the SS stage is grounded AFTER the capacitors, so they are alowed to charge up to the output voltage of the tube stage. At the same time, the SS stage bias current is zero, by virtue of both sides of the bias generator being shorted to ground, i.e. the bias generator is shorted (it won't mind).
When the relay switches over to operating condition, the 10uF cap across the bias circuit sharges up increasing bias to it's normal value (it will take a fraction of a second so no problem), while the twi input caps only change voltage slighthly to accomadate the bias change.
This will work just fine, but of course will not protect the speaker in case something blows up and you get DC at the output!

Oh yes - also, add clamping diodes on the output ('arrows' pointing from output to +rail, and from -rail to output). This also clamps the input voltage to slightly over the SS stage rails, even without a load, although with BJTs damage due to charging coupling input caps is unlikely, it is a cheap precaution, and also helps if the amp is ever muted suddenly while providing considerable output voltage.
 
Administrator
Joined 2004
Paid Member
Hi Owen,
For now you can use terminal strips. Your next incarnation will be more complicated and so will use a PCB. Whatever works for you.

Hi ilimzn,
Thanks, that's what I would have said. I implement the signal relay by simply shorting the lines to ground. Powering up the relay releases the ground. Works like a champ. I also use diodes reverse biased from the supplies to ground. This protects against supply reversal in the event one fuse blows with a shorted output. It then limits that DC to one diode drop. Before the fuse blows, there would be an offset problem. A relay to connect the speakers would be my preferred method.

Yes, reverse biased diodes from the speaker output to the supplies should be a common feature. I do the same thing. In kind, with this hybrid, the same should be done on the input to the output stage.

-Chris
 
Administrator
Joined 2004
Paid Member
Hi Owen,
1N4148's (signal diodes, low capacitance) should work well for the input. 1N5408 (or lower voltage, 3A rectifier) for the speaker and supplies. You can try 1N4007's (1A) and they may be fine. Depends on the amount of energy trying to go where it shouldn't. This one, 1A will be fine thinking about it.

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