diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Solid State (http://www.diyaudio.com/forums/solid-state/)
-   -   Lin topology current driver rebuild (http://www.diyaudio.com/forums/solid-state/80596-lin-topology-current-driver-rebuild.html)

Shredly 31st May 2006 07:05 AM

Lin topology current driver rebuild
 
I have a guitar amp with a Lin topology power amp in it. The push-pull AB-class current drivers are made from TIP31C/32Cs Darlingtoned with MJ15001/002s. I can't find a good source for them, so I am looking at TIP41C/42Cs and 15003/004s as replacements. I believe from several things I have seen that these should be plug-n-play replacements, and I don't think I'll have to re-bias anything. According to the data sheets, the differences are that h(FE) is a bit higher, V(CE)sat is also a little higher-- but since they're biased fairly strongly into the active region, this shouldn't matter-- and V(BE)on is a little higher, which should somewhat offset the current gain difference. I'll be checking to make sure that they're into the active region, and the amp has a current limiter stage just before the push-pull stage. Has anyone had experience with doing something like this and could give me an idea if it will work OK?

I am also going to add a fan as well as some vent holes (I plan to drill vents in the bottom of the chassis and have the fan blow out- this will draw cool air across the power transistors and heat sink). This is where the problem came from in the first place- I played it too loud on a hot day and one of the drivers went intermittent on me. Does anyone know of anything that might help cool the amp more?

The amp has always been a little noisy, and it's definitely the power amp, not the preamp- I have switched jacks and I've tried it with shorting plugs to short the input. Breaking it open I found that the power supply is just a center-tap transformer at 2:1, a bridge rectifier across the ends of the secondary, and a pair of 6800uF caps, and that's it. I am planning to put a regulator on it; there is sufficient voltage to add a pair of diodes to the center tap and use two positive regulators, and I am getting a pair of LM338s with all the trimmings; I have a thread running asking for technical advice on the Power Supplies forum, but I think I've figured it all out, unless someone comes up with some last-minute problem. However, the designers have done some rather peculiar things with ground:
1. The speaker jacks are connected to a network with a pair of 0.68 ohm 10W resistors in parallel to ground, and a "T" with 220 and 22 ohm resistors on the arms and a 0.022uF cap and 10 ohm resistor in parallel to ground; the 22 ohm arm of the "T" goes to the 0.68 ohm resistors. This seems rather odd to me, and I suspect there might be some noise from a ground float as a result, but I also suspect that those 0.68 ohm resistors are a last-ditch effort to handle a shorted output. Does anyone know if this is a good conjecture on my part, and if this arrangement is likely to cause noise?
2. The heat sink is completely insulated from the transistors, but instead of tying it to ground directly, the designers have used a 0.1uF cap. I suspect that this was not a real great idea, and I intend to replace the cap with a jumper. Does anyone know of a reason why they would do this, and whether this could also be a source of noise?
3. I'm looking at the board, and there are- hold on to your hat- three grounds. The one tied into the center tap of the transformer goes to a lug that ties to the chassis, so that's normal- but then it goes out and gets split into two traces (yes, kids, I have two ground traces running side-by-side all the way across the amp board to the connector that goes to the preamp, and yes, both of them go to that connector, and then (and ONLY then) one of them connects to the main power amp ground). After being routed to the far side of the board, "ground" (whatever the heck "ground" means when there are three of them) then goes cruising back across the middle of the board to act as the actual ground that the amplifier uses for all its bypass caps and so forth- and all the way back where it came from, [i]the same freakin trace goes past the place where the original split occurred, and there they are: three "ground" traces running right next to each other. And to top it all off, the schematic has one of the grounds labeled as "dirty ground." GAH!!!. Now it's my experience that ground problems are BAD in audio equipment, and this looks to me like the ground loop from the heart of evil. My tendency is to star ground everything- just go nuts with lugs and heavy-gauge wire. Does anyone know anything about why any sane individual would make a design with multiple different grounds when there isn't any indication that any of them are actually electrically isolated from one another? What am I missing here? Is this voodoo, or is there an actual design concern that I need to learn about here???

Thanks in advance for any help anyone can provide on these questions. If someone needs to see the schematic, I'll try to do an attachment- we'll see if it's actually readable at less than 1000 pixels across.

Shredly 31st May 2006 04:55 PM

2 Attachment(s)
The schematic, for reference.

jaycee 31st May 2006 05:41 PM

The transistors you mention will be suitable for replacement, however you will have to rebias the amp.

The bit you labelled "Current Mirror" Q5, is actually the bias circuit.

Some of the noise might well be the zener diode in the current source feeding the diff amp pair.

darkfenriz 31st May 2006 06:00 PM

Quote:

1. The speaker jacks are connected to a network with a pair of 0.68 ohm 10W resistors in parallel to ground, and a "T" with 220 and 22 ohm resistors on the arms and a 0.022uF cap and 10 ohm resistor in parallel to ground; the 22 ohm arm of the "T" goes to the 0.68 ohm resistors. This seems rather odd to me, and I suspect there might be some noise from a ground float as a result, but I also suspect that those 0.68 ohm resistors are a last-ditch effort to handle a shorted output. Does anyone know if this is a good conjecture on my part, and if this arrangement is likely to cause noise?
These 0.68R resistors are speaker current sensors, so that feedback is partially voltage sense and current sense. This gives the whole amp higher than hi-fi standard output imedenace and lets speaker to resonate at some 75-85Hz.
Proper grounding seems very hard to me here...

Shredly 31st May 2006 07:36 PM

Quote:

Originally posted by jaycee
The transistors you mention will be suitable for replacement, however you will have to rebias the amp.

The bit you labelled "Current Mirror" Q5, is actually the bias circuit.

OK. I assume I'm pushing the V(CE) in the current driver just far enough up into the active region to ensure linear operation, for the one of the four transistors that is the lowest, right? That should be V(CE) just barely above V(CE)sat(max) from the data sheets for the one of the four of Q8-Q11 that has the lowest V(CE) with relation to its particular V(CE)sat(max), if I understand the operation of the amp correctly; that will give the maximum possible loadline for all four transistors, and output will then be limited by the least of, current available from the supply, top of the lowest loadline.

I believe that I should be starting with the pot AP-1 at minimum resistance, which should give minimum current through Q4, and slowly crank the resistance up, checking the V(CE) on Q8-Q11, and watching the current thru Q4 to make sure it doesn't get too high; if it does, then I'll have to consider changing the value of R16 to a higher value, I'll start with 750. If, on the other hand, I can't get the current high enough, then I'll have to move the bias point of Q5 up a little bit by reducing R14 and/or increasing R13, as long as this doesn't push Q4's current too high.

Once I have the correct minimum V(CE) for Q8-Q11, then I want to check the quiescent position of Q4 on its loadline and make sure it's as close as practical to the center of the loadline. If not, adjust R16, R17, and finally R13/R14, and as a final position consider adding a resistor to the collector of Q5 to mirror R17.

Do I have that right?

Quote:

Originally posted by jaycee Some of the noise might well be the zener diode in the current source feeding the diff amp pair.
Ahhh-HA! Yes, of course, I should have seen that. And it's coming in from the +40V supply rail, as ripple. So my project to regulate the power supply is a Good Thing, and if that doesn't take care of it I might want to consider bypassing D3 with a beeper cap, or replacing C4 with something a little more hefty, have I got that right?

Quote:

Originally posted by darkfenriz
These 0.68R resistors are speaker current sensors, so that feedback is partially voltage sense and current sense. This gives the whole amp higher than hi-fi standard output imedenace and lets speaker to resonate at some 75-85Hz.
Proper grounding seems very hard to me here...

Yes, proper grounding seems pretty dicey to me too. My question on this is, do I need to worry about it as a potential noise source, and if so will regulating the power supply help? And is there anything that I might do to it to improve matters without compromising the protection circuit, if it's not enough just to regulate the power?

Regarding the resonance, remembering this is a guitar amp, 75-85Hz is around the open low E string (82.407Hz at concert pitch), making the bass strings on the axe sing nicely. Experience with the amp confirms this; E thru G are quite strong on this amp, with all my axes. G# starts to fade a bit, and A is less strong. 110Hz is, of course, the A string.

Regarding the higher output impedance, again, this is a guitar amp, so it has to handle anywhere from a 4-ohm to a 16-ohm load (one speaker, two speakers in parallel, four speakers in series-parallel, or two speakers in series; or potentially, 16-ohm impedance speakers, which is not uncommon for guitar amp speakers, although it is almost never seen in the hifi world); bear in mind as well that I am running this amp with the two internal speakers in parallel, for a 4-ohm load on the left channel, and with a 2-12 parallel cabinet on the right channel, which is also a 4-ohm load. That might affect the recommendations you give, and if so, I'd be interested to hear about it.

Thank you both for your advice; it is most valuable to me.

darkfenriz 31st May 2006 08:28 PM

I am not sure you took my comment.
I was not talking about load impedance handling, but series source impedance, i.e. very low load damping. While it is recomended to have high damping for hi-fi (damping factor of 30 and more up to thounsends), guitar amps let cone undamped, because resonace is believed helpful.


this might be helpful, although Rod does not say much about grounding in low damping topologies.

best regards
Adam

Shredly 31st May 2006 08:57 PM

Quote:

Originally posted by darkfenriz
I am not sure you took my comment.
I was not talking about load impedance handling, but series source impedance, i.e. very low load damping. While it is recomended to have high damping for hi-fi (damping factor of 30 and more up to thounsends), guitar amps let cone undamped, because resonace is believed helpful.


this might be helpful, although Rod does not say much about grounding in low damping topologies.

best regards
Adam

Ahhh, I see, yes, you are right, I did misunderstand you. Yes, in fact, resonance is very important to the sound of a guitar amplifier. If you listen to Carlos Santana on his early albums, playing with a hollow-body guitar and a compander, and getting resonances (limited feedback) during his leads, you'll understand precisely why. The upshot of this is I probably need to not fool with it, because it will remove some of the sound I like.

I'll review that article in more detail and analyze the design a bit more, too, so that I comprehend what the designers were trying to accomplish. Perhaps it's not just a matter of safety feedback.

Thanks very much. :D

Shredly 1st June 2006 06:31 PM

Jaycee, or anyone else with knowledge of biasing Lin amps, I'd really value some feedback on my proposed biasing steps if you would please take the time to look them over.

teemuk 1st June 2006 08:08 PM

Concerning regulation, it's better to regulate only the differential stage supply and perhaps the VAS supply. There's a constant current source at differential but it seems to be highly dominated by the fluctuations in Vee. If you regulate the differential supplies the circuit should become less sensitive to ripple and effects of supply sagging. I guess this is one of the main sources of the noise you are experiencing.

Your explanation of the grounding still remained unclear to me. A picture might be much more visualizing. As far as i know it's quite common to run separate power and signal grounds. If there are digital chips in the circuit they should also use a "separate" ground (which is so full of transient switching currents that it could be considered "dirty"). Maybe this is the case.

Concerning the current sensing circuit; I did some simulations with different feedback methods about a month or two ago and wrote about the results to a thread at ssguitar.com.
http://www.ssguitar.com/index.php?to....msg785#msg785
Reading it might be a good continuation to the Rod Elliott article, just don't get repelled because of talk about output transformers. You have to register to view the attached pictures though.

The current sensing / mixed-mode feedback circuit is a quite common feature of guitar amps probably because it "mimicks the tube amp response". Actually the circuit has nothing to do with mimicking tube circuits, both circuits just tend to response to load impedance similarily. I guess the high current ground path of the circuit should be grounded the same way a speaker load usually is - which is straight to the main star ground point. The low current grounds of it could go to signal ground, or so I guess. Anyway, most of the grounding theory is just speculation since it's highly circuit/PCB design/layout etc. related. You have to test what works best.

Shredly 3rd June 2006 10:07 AM

Thanks, that was helpful.

The transistors showed up today, and it looks like they were plug-n-play after all. I'm going to do more with it tomorrow, but at this point, it looks like the regulators removed the noise and the adjustments needed for biasing were minimal. I made two modifications besides changing the output transistors and adding the regulators:
1. The heat sink was bypassed to ground with a .1uF cap. I replaced this with a jumper.
2. The ground from the service was attached to the case, but the ground from the amp was not. I attached the ground from the amp to the case where the power enters the Lin amp.


All times are GMT. The time now is 08:36 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2