Adjusting Vbias on one's amplifier

[JohanK]

My brother (he's the leader, I'm following) and I were attempting to adjust the Vbias on our respective Adcom amps; his is the 555II and mine is the 535II. I read that the Vbias in a complementary feedback bipolar amp should be 8 mV for optimum sonics (from book by G. Randy Sloan).

My amp had 12.8 mV on one channel and 11.3 mV on the other. I got the Vbias set to 8 mV on each channel with no load and with the heatsinks at a warm temp (I left the amp powered on without speakers or interconnects for approx. 1/2 hour). The amp produces more SPLs, is less harsh (lower THD in upper frequencies?) and seems to have more bass (all subjective conclusions, no data to back it up).

Is this the correct proceedure (i.e. is the 8 mV measurement a 'quiescient' measure)? Or does the amp have to be as 'hot' as it would be under normal usage? Is 8 mV the correct figure I should be shooting for?

Thanx.

[JohanK]

Jon, can you help?

[ThomasW]

Johan

I sent your post to Jon. Verbally he told me that without special (meaning expensive) measurement tools/test equipment it was impossible to accurately set the bias. And that theories about a "generic" bias setting are flawed. The bias should only be set to the factory standards.

Jon is currently swamped at work. If he gets some free time he'll try to reply with more specific details




theAudioWorx
Klone-Audio

[JohanK]

Thomas,

Thanx for your efforts. I think I'll contact Adcom in hopes of obtaining a service manual.

Johan

[JohanK]

I posted this same question at AVS b/c Nelson Pass is the special guest this week. Here is his reply for those who are interested:

JohanK,

Setting bias is simple; put it at the best sounding setting.

Ok, Ok, I know that's not helpful, but it is difficult to generalize. First off, you don't set bias voltages, you set bias current, and measurement of voltage across the Emitter / Source
Cathode power resistor is your indicator of the bias current. The bias current is the voltage divided by the resistance; for example 10 mV across a .33 ohm resistor is 33 mA of current.

Most of the time the temptation is to set the bias at as high a current figure as the amplifier can stand. This works great for Mosfets, which get more linear, sound better, and measure better
as the bias is increased. I try to run the heat sinks around 50 deg C, which is the point at which you can only touch them for a few seconds.

With Bipolar output devices, the highest bias setting does not always improve the measured performance, in fact it sometimes degrades it,
but I usually find that it sounds better there anyway. Most bipolar AB amps are run pretty cold, but from a measurement standpoint they usually work well enough with 100 mA bias total.

Keep in mind that bias adjustment must be done very carefully - it is easy to lunch an output stage with some carelessness. Consider using lower AC line fuses and a Variac to run the line low as you make your adjustments while monitoring the current through the output stage. Confirm that the current sharing is roughly equal among parallel output transistors. Readjust the bias figures over the course of an hour or two and operate the amp with the lid on, as air currents can affect the results.

[JonMarsh]

Well, I'll put my two cents in here, too, though Nelson has obviously covered the bases pretty well.

For AB bipolar amps, using a simple emitter follower, or some variant on it, such as an emitter follower triple, there's an optimum bias point at a low current value, depending somewhat on the choice of emitter ballasting resistors.

If you put a distortion analyzer on the output, and start with the bias turned down all the way, you'll see high order distortion products due to the crossover notch, the area where neither set of output transistors is conducting. As you increase the bias current gradually, the distortion will drop until you hit the point where the positive and negative transistor sets smoothly overlap their turn-on/turn-off points. This will be a null on the distortion analyzer output.l If you continue increasing the bias, you'll actually have some conduction overlap, and the transfer function of the output stage will not be as linear, and the distortion will increase some, though not nearly to the levels at the bias point "below" the null point. If you continue increasing bias sufficiently, then you will be operating in small signal class A up to a number of watts output, and the distortion will gradually fall to a fairly low value, though it may not be as low as at the "null". However, the distortion products in this mode will be fairly low order, which is quite benign.

The danger in all of this is the old bugaboo of thermal runaway. As bipolar transistors, or vertical DMOS transistors, for that matter, heat up, the junction potential or threshold voltage drops, and the output quiescent current increases. Which makes them hotter. Which increases the current. Etc, Etc., until things are crispy critter. That is surmounted by using a thermal feedback element which reduces the bias as the output device temperature increases, in order to maintain bias current stability.

MOSFET's used in audio Amplifiers are often a different type of animal than the vertical DMOS transistors made by IR, Infineon, Motorola, Fairchild, etc. They are lateral MOS devices; they have higher on state resistance, but fairly high thermal stability; the bias current "Q" point for stability is about 200 mA. For a given reasonable bias voltage, they'll tend to warm up to that current, and stop and stay there, because of intrisic mechanisms, mainly temperature related affects for transconductance, channel resistance, and threshold voltage.

MOSFET's aren't ideal devices, though, because they have relatively low transconductance. That means it takes a fair amount of voltage change in the "goes into" to produce much of a current change in the "goes outa". This means that a simple MOSFET output stage, configured as a complementary follower, has a fair amount of distortion open loop.

Which brings us to another output stage topology a few manufacturers use- compound output stages.
What is a compound output stage? It's an output stage which uses drivers and output devices configured in a local feedback loop to increase linearity and reduce distortion, making the stage less dependent on loop feedback and input biasing for high linearity. Bryston uses this technigue with their bipolar amps, and I've used it for years (decades) with MOSFETS, employing a class A driver and AB outputs in a compound configuration. This results in an output stage with excellent linearity, less sensitivity to biasing, and low output impedance open loop, all hallmarks of a better sounding circuit.

Nelson Pass's X-Series is also a type of compound output stage, configured for some special properties in a balanced configuration. The patent is very interesting reading.

So, regarding your Adcoms, if you want them "teaked to the nines", borrow or rent a distortion analyzer, or find a buddy nearby that has one. (yes, I know Crown and HP disotortion analyzers are *not* normal home appliances- HT in general still isn't "normal", either!). The advantage to using the analyzer, if you can borrow one, is that you can do the setup much more quickly and easilty than trying to gauge by ear. Last, keep in mind that whatever spec Adcom uses is *most likely* based on the voltage required to hit that null, so this may be mostly an intellecutal and learning exercise- which is not a bad thing, but may not change the way your system sounds from before hand.

Regards,

Jon




Earth First!
_______________________________
We'll screw up the other planets later....

[JohanK]

Thanx for the info Jon. I appreciate you sharing your knowledge.

Regards,

Johan

P.S. If one were interested in building an amplifier, are there any good (bipolar) amp kits available? What electronics store(s) do you recommend for obtaining amplifier parts; i.e. emitter resistors, output transistors (Motorola in particular), power supplies, heatsinks, metal oxide resistors, etc.? Thanx.

[JonMarsh]

There used to be a variety of options in kits, like, 15-20 years ago. Vellemen, from England, has kits, but since you mentioned "good", I'd have to probably rule them out- they're a bit touchy, and from reviews, not completely stable with all load configurations.

Marchand has some nice kits with all MOSFET designs; a relatively straightforward two stage differential amp, and complementary output stage. All amplifiying devices are MOSFETs. The circuit and implementation look good on paper, but I haven't heard them.

I must confess to not following kit availability in detail, for the obvious reasons. When I want to build, I build from scratch, with my own designs and PCB. When I want to buy, I buy Aragon. They're bipolar, and one of the best values in commerical amplifiers, in my opionion, regarding cost versus total finesse in electronics. If I were on a budget, I'd own Aragon 8002's. Since I've been "lucky in love" with finding used amps, I own a pair of Palladiums. If I weren't working on a CLASS D amp for the bottom end of my system, I'd probably pick up an 8002 for that. (The "bass bins" in my X1 clones are about 96 dB efficient, in the range they're used (25-150 Hz), so amplifier power is *not* an issue, as long as it's clean and very well controlled.)

Buying parts for amplifiers isn't super hard, but it isn't super easy, either.

Many general purpose electornics distributors like Newark (now owned by Farnell), Digikey, Mouser, etc.,will take online orders through a credit card, so it's pretty easy to do business with them.

Just for you HTGuide Guys, here's who I use for what for many things:

Plitron, Canada- Toroidal power transformers and cut core inductors (yeah, I'm Mr. Inductor input power supplies- you wouldn't believe how much cleaner that makes the unregulated DC bus, *and* the AC line currents, compared with a convential capacitor input supply). Ayre Acoustics, Cello, and that Brit company handled by Audio Advisor also do that.

Newark: Lots of general purpose Motorola/On Semi parts, IRC resistors (use metal film 7 watt power resistors in driver/output sections), AD/PMI parts like MAT-01, MAT-02, MAT-03. AD video power buffers; use them as line buffers in elec. crossovers, preamps. GE Polypropylene caps.

Digikey: Silver mica caps, Polypropylene small value caps, Panasonic Electrolytic power caps. Burr Brown Op Amps (OPA627, OPA134, OPA604, BUF634)

Mouser: Dale/Visha metal film resistors, switchcraft and Neutrick connectors.

Welborne Labs: Cardas Connectors.

Farnell: Heatsinks, power resistors, Magnatek Lateral Power MOSFETS, from England.

Regards,

Jon




Earth First!
_______________________________
We'll screw up the other planets later....

[JohanK]

Jon, thanx so much for your help. I hope I can continue to ask questions here as the need arises. Thanx again.