AR Amp Bias

[Mexicomike]

To follow up on this topic of a while back, I offer this info re biasing/modding the AR amplifier. This is a result of working with it over the past several weeks and doing a lot of research and testing on my amp. First I have to again say I am NOT an electronic circuit expert BUT I have learned a lot over the past month or two from studying and email discussions and now know a lot more about this subject than I ever really wanted to!

First, as I posted previously, I preferred not doing the factory mod to my recently-acquired original pot-equipped amp. This was for two reasons, I didn't want to alter the amp and I believed that the original AR design (bias pots as opposed to the later factory mod of fixed resisters) was a superior design. True, in operation the original bias pots were of rather poor quality and caused a lot of overheating problems. In fact, I personally owned 3 of the AR amps back when they were produced - the first two were replaced by AR for overheating (one caught fire!) and I sold the third one soon after receiving it from AR because I didn't trust it!

Anyway, as I posted, on my present AR amp I did a LOT of cleaning of the oem bias pots and did get the amp to be quite well behaved and stable but still could not keep the bias current and offset from driftng considerably in operation, though not enough to cause any heating problems. But I wanted the amp to be as good as it COULD be. So I decided that better pots were in order.

I purchased 4 Bourns 50 Ohm pots http://www.bourns.com/pdfs/3299.pdf and some additional 33 Ohm 1/2 W shunt resistors. I could have used the 33 Ohm resistors that are installed across the oem pots but based on how I chose to install the new pots and the fact that I wanted virtually NO changes to the amp topology, I left them in place. The first pic shows the oem layout of the amp; the second shows a closeup of the oem bias pots in one channel with the oem resistor across the pots. The third shows the new pots/resistors installed (glue) on top of the old pots which are no longer in the circuit. Since the last pic was taken, I cleaned up the leads so the installation is neater now than the pic shows though the new pots are in the same spot. All the original amp parts remain in place. To restore the amp to absolutely stock configuration - as it left the factory - requires only the removal of the new pots/wires and resoldering 1 oem lead to each of the oem pots.

Originally I decided to set all the pots to the resistance specified in AR's factory mod for fixed resistors and check the results. THis would simulate the factory's fixed resistor modification. I tested with both 12 and 13 Ohms, which AR says are the suitable resistor replacements. Obviously, they produce different bias current and equally obviously, either current SHOULD be within the operating range of the transistors according to AR. This is one spec that I have never been able to find - the suggested bias current.

The problem with the fixed resistiors is that even though I selected specific 12 and 13 ohm resistors that tested within .1 ohm of their spec, the bias current was not the same for each transistor and the resulting DC offset was not all that impressive because of circuit variations. Offset was never any better than 70mv. This is within the AR factory mod specs but, according to everything I have been able to find re amplifier offset, it is rather poor. As an example, the specs on my Moscode 600 call for 0 DC offset with 20mv being the maximum acceptable offset. So it was clear to me that the AR factory mod was a CHEAP way for AR to fix the overheating problem but their original design - given good quality pots - could produce a BETTER amplifier.

So now I used the adjustments on the pots to tweak the readings. This required different resistance settings on all pots. But with the pots, the DC offset is now within a stable 2mv on both channels - a huge improvement on the factory fixed resistor mod. It also allows you to vary the bias current (usually increasing it over oem) to produce optimum amplifier performance. This requires either some test equipment or a lot of listening tests and attention to the heat sinks as higher bias current (to a point) will improve performance but increase heat. Suggested bias current in a variety of sources suggests anything from 25 to 150ma. The current bias current (pun intended) for my amp is 55ma which produces just slightly warm heatsinks after 1/2 hour of no-signal operation. Since I do not have the appropriate test equipment, I can only perform listening tests and I hear a difference in the sound between the present setting and the previous settings and between the present settings and the fixed settings. The primary difference that I notice is that there was some harshness in the sound in certain passages previously that is not there now. The sound seems more "solid" and coherent - less "transistory" in the old sense of how transistors often sounded.

So my recommendation re these amps is NOT to perform the factory fixed resistor mod. You can do much better by installing quality pots and achieving what the AR amp was intended to do in its original design, not what someone decreed as the cheapest way to deal with the overheating! If I had an AR amp with the factory mod, and I wanted the amp to perform to it's maximum capability, I'd remove the fixed resistors from the circuit (but leave them in place) and put in the pots!

[mrdb]

Nice job. I'm glad that everything is working ok now. I have a question and an observation: First, what is the wattage rating of the Bourns pots? Also, I notice in your overall view showing the four pots that the diode attached to the rightmost power resistor below the heatsink is not centered in the blue glue. Is the diode in contact with the resistor? It must be for thermal tracking of the bias. It is not clear from the picture if it is touching. The glue could be some thermally conductive adhesive or it could just be there to hold the diode in place during transport.

As you said, you learned a lot more about the amp than you wanted to know. The simple series output circuit is very sensitive to component drift. The amount of bias current through each half of the circuit determines the quiescent conduction of each transistor and any imbalance in currents introduces an offset voltage at the speaker terminal. The beta (gain) of the drivers and outputs affect this as well, and varies with temperature. I don't think that having vanishingly low offset voltage is really critical, and anything under 20-30mv would be fine, although zero is certainly an ideal goal. I suspect that you will find some drift in offset voltage from a cold amp to a very warm one. The newest amps use a servo circuit to maintain very low offsets, but the circuitry is much more involved. The amount of power dissipation in the speaker from the offset voltage is: offset voltage squared divided by the speaker dc resistance, which in a case of 100mv offset and 3 ohms (the typical dc resistance of an 8 ohm speaker) would be 3.3 milliwatts, certainly not any kind of problem.

Again, congratulations on a job well done.

Bob

[Mexicomike]

Thanks Bob,

The 2 diodes you noted have, in fact, popped out of the blue "glue." I have ensured they are touching the resistors but would like to reglue them with some appropriate adhesive but I'm not sure what that would be. The blue feels like an epoxy of some sort but may not be. It's extremely hard and brittle.

The new pots are 1/2 Watt, the same rating as the resistors AR recommends to replace the oem pots.

As you noted, the current and offset does vary from a cold to warm amp, though the offset doesn't change too dramatically. If I adjust the offset to 0 when first turned on, it will end up around 15-20 mv when warm. So I set it when the amp has been on for about 30 minutes. I did that yesterday and right now - I'm still monitoring readings and haven't closed the amp back up yet - the offset is averaging around 5 mv.

I have not replaced the filter capacitors which are original. I don't have a clue if they affect any of this. If my understanding is correct, bad filter caps produce hum. Since the amp produces no hum, I'm assuming the caps are OK. I have 2 new filter caps and even if the old are OK it would make sense to replace them now with the amp open anyway. If I replace the caps I will keep the old ones so the amp can be returned to "original" just as with the oem pots.

Next project is to replace the Caps in the AR3as as per the AR3a paper to ensure I don't damage the mid and high range drivers with the rejuvinated AR amp!

[mrdb]

Mike,

Most service literature recommends adjusting bias and offset after the equipment has been in operation for 15-30 minutes. With your new pots and the fixed parallel resistors, everything should be more stable. Regarding the filter caps: the amp design is nearing 39 years old and if the caps are original they are getting pretty tired. I had to replace one in my AR amp just to keep it going temporarily. I was using it for the surround channels in my home theater system. Now that I have a different amp in place, I will replace all of the filter caps in the AR, and maybe the caps in the audio path in addition. My amp started humming so I knew where to look. I have an oscilloscope so I was able to look for ripple voltage on the power supply lines. Besides the main 5000 microfarad caps, C49 and C50, I would recommend replacing the two caps in the idler supply, C56 and the three-section cap C31 (which is also used in the 20 volt requlator. While you are at it, consider replacing the dual 2000 microfarad cap used in the driver section. I am referencing part locators from the amp schematic in the archive section. If you are a glutton for punishment, replace all of the small electrolytics, too. If you do, the amp should be good for another 30 years!

Bob

[mrdb]

Mike,

I opened up my amp last night and re-did the bias of the outputs. It was tricky. I wish that I had some 50-ohm pots! Anyway. I got one channel offset at 20mv after 30 minutes of operation and the other channel at about 70mv. The bias current is 50ma on one channel and 60ma on the other. The heatsinks run lukewarm after half an hour of playing music at a low level. I checked harmonic distortion at about one watt output and it was 0.1% on each channel. I looked at the distortion analyzer's null (distortion) output on a scope and did not see any evidence of crossover notch distortion at low level.

Regarding changing the electrolytic caps: it looks like some of them require some disassembly to get to them. The most important ones are the two main filter caps. One is easy to access, but it looks like the tone board must be removed to access the other ones. In my amp, luckily, the more easily accessed one was bad. I think I will leave the other one alone as long as it works. Unfortunately, when one goes, the others are usually not too far behind. I replaced the electrolytics on the driver board since it was easy to remove.

Like you, I noticed a smoother sound after re-biasing. The idle current was much too low before (less than 7ma) and the amp sounded harsh, probably because of crossover distortion. Unfortunately, I did not check distortion before I re-biased.

Bob

[Mexicomike]

Bob,

Assuming all the AR amps are constructed the same way: When you remove the top there are three screws holding the forward portion of the top to the front plate assembly. There are three additional screws on the bottom front of the amp that also hold the front plate assembly. If you remove these screws the front assembly can be moved downward (sort of pivoted) and out of the way - there is sufficient wire slack to allow that and the board attached to the front plate goes with it. Nothing needs to be disconnected. Then the access to the big caps is fairly easy.

I am still tickled to death with the results of the re-pot. The amp been playing quite loud all morning and the heat sinks are warm but not hot. I just rechecked offset - same reading as yesterday, around 5mv!

I decided to replace all 4 of the big electrolytics but I don't have their replacements at the moment and won't be able to get them until I'm visiting in the States in March. So I think I'll button it back up and call it done for now.