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.
Administrator
Joined 2004
Paid Member
Hi Owen,
Your bias circuit may not be set up properly yet. Measure the B-E voltages on your outputs (don't slip with the probes!!). If you are not getting enough voltage to bias the outputs on, decrease R9 to 1K and try again. You will need about 2.3 V across Q3 C-E.

-Chris
 
VERY IMPORTANT! YOUR Q3 MUST BE ON THE HEATSINK WITH THE OUTPUT TRANSISTORS!!!, or giving them even a small bias current will result in them going BANG, it's just a matter of time - due to thermal runaway (not to go into details here). If possible, keep the wires to Q3 short, if not, you may need to move the 10uF capacitor across C-E of Q3 right onto the pins of Q3.

The reason why it sounds bad at low volume is nonexisting or insufficient bias. This is also the reason why bootstrapping worsens the sound - bootstrap is a form of positive feedback, so if the whole SS stage distorts without it (because of the low bias) it will make things worse before it makes them better.

The 220k to ground (two of them) should be from each base of your BD139/140 drivers to ground, so, looking from the input, after the 2.2uF caps. You don't really need anything from the input to ground as such, not in the finished circuit, though for experimenting with directly feeding the SS stage with a CD player and such, you may want to add another resistor to ground (220k will do) - but for the tube front end, you will have to remove that one.

As Chris says, the load the tube stage will see, will be half of 220k, so 110k, but only with bootstrap. Without it, it will be about 10.8k, as your 4x 12k resistors to the power rails also figure in this equation. Bootstrapping makes them 'invisible', of course, once bias is set properly as I said above.
 
hey chris,
I did as you said to and increased the bias to 6.6mv accross botht he resistors and it improved a LOT.. i increased it further to 8.5mv (almost the max on the pot now) and i could almost live with the amp now.. ALMOST..
It works great at low volumes aswell now since i added the two resistors back in (i used 270K because i couldnt find any 220K ones), still havent added the zobel net again or the bootstrap caps or the input cap... but it deffinatley sounds good.. DC offset is about 20mV with load and about 25mV without a load now aswell.. there is still a slight buzz when its first turned on but i think thats because i havent got a delay on the HT on the tube stage so HT is applied with heaters cold still :bigeyes:

So to summerise what i still have to do now:
1) fit zoble cap and resistor from output to ground
2) small cap from input to ground
3) tweak bias network so pot isnt so twitchy and starts adjusting bias before it is almost at the end of its travel
4) make soft start circuit for tube stage

Does this sound about right or are there other things i should be doing first?
Many thanks,
Owen
 
we posted at the same time ilimzn! Q3 is on the heatsink with the output transistors. I will move the 270K resistors directly to the transistor legs? or are they ok from the SS side of the 2.2uF capacitors to ground? and i will try it with the bootstrap capacitors in place again. I t is actually being driven by the tube stage and is soundign good at all volumes now.. it still needs a bit of work to make it perfect but i am very impressed especially seeing as it is on a breadboard still! lol
Owen
 
270k placement is not overly crytical as long as they conenct to the same electrical point.
Be careful when optimizing the b ias pot part of the circuit - you can get the pot to react nearer to the middle by reducing the resistor it has in series, just make sure that you can turn bias completely off. Also, keep in mind that bias increase and ot travel are not proportional, bias rises a lot faster.
My guess would be that you will get a sweet spot regarding sound with about 20-30mV measured over both resistors (so half that on each single resistor).
Using a bootstrap arrangement will also make it MUCH easyer (ovr 10 times easyer, to be precise) for your tube front end to drive, and it will also improve the SS power supply noise rejection for the SS stage - this will also improve the sound. Keep in mind that crossover distortion of the SS stage due to low bias will be exaggerated by bootstrapping, but proper biassing will reduce distortion by unloading the tube stage. This means that there is a 'sweet spot' involved. You can also vary that 'sweet spot' by applying partial bootstrapping - this is done by using a resistor in series with each bootstrap cap. If the resistor is zero (short) i.e. only caps are used, in theory, it makes the 12k resistors to ther ails 'invisible', i.e. multiplies their value to near infinity. On the oher hand, 1/10th of the 12k resistor value, so 1.2k, will multiply the resistance of these 12k as seen by the tube stage, by a factor of 10 (so you can easily see the relationship here). Have fun playing! ;)
 
Administrator
Joined 2004
Paid Member
Hi Owen,
Everything ilimzn said I agree with. Decrease R9 to about 1K and try it again, remember to turn the pot back down to about 1/2 way first.

Try to resist "the hotter, the better" nonsense. Just have your wife or a child (free from mental nonsense) listen as you adjust the bias settings. Same for comparing CCS with bootstrap. Distortion measurements will help keep you "real". At some point, try with another stage. This should vastly improve high level performance.

-Chris
 
ok.. little update, i added the zoble stuff ont eh output again so that is ok.
added the cap ont he input but the amp REALLY didnt like that.. took it off and its all good again..
I'm going to shorten the wires on the output devices and on the bias transistor and move the bias transistor so it is between the two output decives ( its currently monuted on a hole that was already in the heatsink)..
i think i will leave the bias circuit at the moment... If i can get the bias to where i want it as it is theres no need to change it.. just need to wait and see.
i juess i should probably start designing a PCB for it.. although it is simple enough that i could probably wire it P2P.. but i guess for SS stuff a PCB is a better idea? bearing in mind there is hardly anythign going onto the PCB...
Oh yeh, need to sort DC offset out aswell.. add the pot to adjust DC offset i mean.. at the moment it is only about 30mV so not too bad.. but i would like to try and get it down a bit still..
need a relay on the output to disconect the speaker on power up as i have a feelign there will be a nasty thud otherwise if i remove the output cap..
so yeh.. just some tweaking left now, happy its finally working as it should be :)
Thank you all for helping me get this far. I know you must all be getting pretty fed up with me by now..lol..
Thanks,
OWen
 
I will still try adding an extra stage sometime.. and i will try useing CCS.. but i think i will just tweak this one a little more and build it as it is.. that way when i build the next one with an added stage i have somethign to compare it with directly side by side.. and i actually quite like the sound i am getting out of this amp at the moment, even though its running without bootstrap and still at a very low bias.. things can only get better! or go bang knowing my luck with silicon..lol.
Many thanks,
Owen

EDIT: next one i build will have a better power supply that way aswell... bigger transformer for output section and a smaller transformer for the tube section..
 
Administrator
Joined 2004
Paid Member
Hi Owen,
Don't worry one bit about 30mV. Ignore that completely.

Don't design the PCB yet, bias is fine as long as it's controlling the bias current. You still need to play with high level performance (extra stage) and CCS or bootstrap. Take your time.

Once you are done, you can design the board and troubleshoot that. ;)

The relay between the tube stage and bipolar stage is a must. Simple timer circuit. Just time how long it takes on average for the plate voltages to settle in and add a bit of time.

-Chris
 
Administrator
Joined 2004
Paid Member
Hi Al,
Why the devil would you subject the BJT's to that voltage spike? It's 100% avoidable.

I have working hybrid designs as well. Mine do not require an output relay, except for protection purposes maybe. There is the slightest of clicks with your ear to the speaker at turn on with only the tube stage to output relay.

-Chris

Edit: Al, there is only an ad for your amp on your link, no circuit details. The link is therefore not helpful except as an ad.
 
Hi AAK, I cant really take credit for the design, If it wasnt for Everyone here I would never have got it working. But I am glad i stuck at it, Now that i know silicon dosn't hate me as much as i thought it did i feel more confident and am looking forward to improveing the design further. At the moment though i want to build it like it is so that i can listen to some music through it for a few weeks and get used to it and then i will be more able to tell if the changes i make to the design ont he next one i make have made any improvements over this design, I have a feelign that when i have sorted all the cabeling out and have set the bias up properly it will sound even better than it does now!
Many thanks,
Owen
 
I found another bug... if i set the bias higher than 8mV (accross both resistors) it slowly but surley keeps riseing all by itself :eek: i have checked that the output transistors and Q3 are makeing proper contact with the heatsink and thats ok.. they dont measure atall above room temperature so cant be heat related anyway as far as im aware? :S mabey i need to move the 10uF cap closer to the bias transistor? or is this totally unrelated? i have just been listening to it pretty much since my last post at 8mV with the bootstrap caps in place and it didnt move a bit.. and sounded pretty good too.. no real difference in soud with and without bootstrap caps at this bias setting but i cant safely get the bias higher at the moment to test it at higher bias settings.. any ideas?
Thanks,
Owen
 
Administrator
Joined 2004
Paid Member
Hi Owen,
Mount the drivers to the heatsink as well. You may see the bias rise but it will (or should) start to come back down after a while.

If you see the bias start to rise, and the rate is accelerating, that's thermal runaway. Not your friend. Power down.

-Chris
 
DoomPixie said:
I found another bug... if i set the bias higher than 8mV (accross both resistors) it slowly but surley keeps riseing all by itself :eek:

Looks like thermal instability. Could be runaway, in that case like Chris said, put drivers on heatsink as well.
It is normal that the bias starts rising on it's own though it should stabilize or even go down somewhat long term. Make sure Q3 is in good thermal contact with the heatsink as well (proper thermal grease used and all).

i have checked that the output transistors and Q3 are makeing proper contact with the heatsink and thats ok.. they dont measure at all above room temperature so cant be heat related anyway as far as im aware?

Actually, it is, because the internals of the transistors heat up quicker than the heatsink, it takes a while for the temperatures to equalize (if you will, it takes a while for the heat to get from the drivers/outputs to Q3 via the heatsink).
To test this you can connect your voltmeter across the 0.33ohm resistors, and slowly add bias to get into the zone where it rises on it's own, say 10mV. Then leave the adjustment alone, keep your finger on the power switch, and just watch. If the voltage ever goes over say 120mV (this would be about 200mA bias current), switch it off. It is safe to say you have a thermal runaway problem. In the circuit you are using, as Chris mentione, it is likely because your driver transistors are not on the same heatsink, so they heat up 'without Q3 knowing about it' so to say. The cure is to put them onto the heatsink with Q3.
If you can get the current to stabilize, then you can go on and push the bias up till it settles at 15-30mV as mentioned before. It is fairly normal to see fluctuation when the amp is switched on and the heatsink is cold. There is a trick to get the process quicker so you don't have to wait (but you have to be careful with that power switch because if there is runaway, the current grows exponentially and it may become TOO quick!), and that is to use a hair dryer to heat up the heatsink (seriously!).

I have just been listening to it pretty much since my last post at 8mV with the bootstrap caps in place and it didnt move a bit.. and sounded pretty good too.. no real difference in soud with and without bootstrap caps

The bootstrapp will show it's advantages at higher volume (or lower sensitivity speakers), and complex loads where the tube section may run out of current. It would help if you could post a schematic for your tube stage.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.