Sansui AU-alpha907DR quiesent current

[eradoncic]

This model of amp is TOTL model in 1992. It is in as new condition, or at least VERY seldom used - does not need even usual cleaning.
However, the sound is a bit veiled, dark, and has especially too hard/strong bass...

I opened the amp and measured inside; the DC offset was very good, the bias (quiescent current) also seemingly OK (7-8mV accross 0R22, which is around 31-36mA per transistor). I dont have the service manual (it is extremly difficult to find for now), so I don't know if the values are OK.

Now things becomes interesting: searching here and audioasylum web site among others, I actually found very little valuable/practical information about biasing the SS amps, but generally lot of people recommended increasing the quiescent current to some 65-70mA, or even 100 mA per device, and others argue that the correct bias can not be set properly without spectrum analyser, etc.

During several days I increased the quiescent current to some 65mA and the sound changed dramatically - more space, better low frequency, smoother sound, in one word - more controlled overall sound. However, I left it around 40mA, just to be safe during resolving this question.

Still others suggests that the voltage accross the Re is of importance, rather than current, and this should be set to some to be 20-30mV, irrespective of emitter resistor value. For the amp in question, the current will be between 68-112mA/per device. The heatsinks of the amp are fairly big, but the amp has 8OT per channel, each 125w derating (C3284/A1303)

I dont know what to do - a higher bias than per SM yiealds subjectivelly etter sound, but on the other hand, why they did not set those value at the factory, or noted on the SMs?

So, what are your opinions and/or experiences

[eradoncic]

Nobody?

OK; let's put the question on the other direction:

For bias adjustment the service manuals of similar models/output says that:
1. preheat the amp with 20kHz sine wave, one third of the nominal power (that's around 60W), for at least 5 min
2. adjust the voltage accross Re (0,22R) to be 8,5mV+/-1mv.
(at this time, the transistrors and heatsinks are fairly hot, I cannot touch them for long)

I've done that, and after turning off, cooling and then turning again on, the bias voltage (and perhaps current too) is going up, to some 16-18mV, and then lowering again but never at initial values of 8,5mV. I understand the tempereature positive correction, but, why then it is advised to adjust hot amp.


Now, the questions are:
- it is then this kind of adjusment actually the bias below which the amp should'nt go even when when stressed?
- what should be the upper bias then (idle current??)

The bias transistor is 2SA1845, as I can see.

Please any oppinion is welcome. EchoWars?

[anatech]

Hi Eddie,
I've been in the service industry for years, so I'll try to give you some of my own biased thoughts.

The original biasing is a level set to maintain the specs for distortion with every unit they ship. The bias could be lower and still meet specs.

A higher bias current causes a higher ripple current in the filter caps, that shortens their life. The same holds true for the power transformer.

A bigger heatsink (to the proper size) may change your temperature coefficient to a negative value, or stable. It's a fair bet the power transformer, filter caps and heatsinks are a little undersized.

Do not ever run an amplifier at sustantial power levels above 2 KHz (some manuals may say 5 KHz). You will overheat your zoble network. If you damage it, the amp may start to oscillate. Not what you want.

Most bias setting procedures are attempting to save the tech some time. Generally they want to get the heatsink hot, then let it cool with no signal (still running) until the bias is stable. You set it to the desired value at that point and let it sit until it's stable again. Readjust and let it settle down.

-Chris

[eradoncic]

Originally posted by anatech

A higher bias current causes a higher ripple current in the filter caps, that shortens their life. The same holds true for the power transformer.

I can understand the impact on the filter caps, but it is not clear to me for the transformer, but anyway, I get the point

A bigger heatsink (to the proper size) may change your temperature coefficient to a negative value, or stable. It's a fair bet the power transformer, filter caps and heatsinks are a little undersized.

Well, this is interesting, really. I always thought (based on the many readings here), that bias is always temperature dependent but positively regulatied. In my amp, after heating, the bias is going to lower, and stabilise at some values. I guess it is very important this value not to be too low or zero.

The interesting thing is also with the power supply: the transformer is really big (I don't know its ratings exept the power consumpion - it is rated 1250W).
I can say it is first class unit (made by Tamura/Sansui), sealed, looks as it is "C" or "R" core, dual secondary for power amp, obviously not cheap product...but, in the same time, for each channel has only two caps: the bigger 5600uF/85V (Nichicon highest grade, KG) and the smaller of only 3900uF/85V (Elna Silmic). Both are almost as 0,33L beer cans . Seems to be a bit small ratings overall, but I think those cannot be inexpensive, and the savings are out in question here (the amp weights 33kg alone). Someone told me that's because of high quality/regulated trasformer used.
I attached the pics of the supply

Most bias setting procedures are attempting to save the tech some time. Generally they want to get the heatsink hot, then let it cool with no signal (still running) until the bias is stable. You set it to the desired value at that point and let it sit until it's stable again. Readjust and let it settle down.

-Chris [/B]

So, you thing it is OK to preheat the amp with some music (mono from tuner, for example), and then to adjust. This sound good to me

[anatech]

Hi Eddie,
Music would be fine. I use dummy loads because I'd like to hear myself think.

The transformer core will run hotter in a quiesent state along with the filter caps. Quality of these components is held equal and the cooler they run, the better. Just compare their temperatures between the two bias settings.

Some Marantz amps have a slightly negative temperature co-efficient for bias current. Ambient temperature rise will affect this for the same reason heatsink efficiency does. I've seen this happen as I move amp prototypes from large heatsinks to smaller ones.

This is another reason why amplifiers shout be able to have good airflow.

-Chris

[eradoncic]

Chris,

You cleared to me lot of things now.
The main puzzle was the reason of preheating and adjusment in such way, and in the same time I found only "conservative" way of setting.

I edited a bit the previous post in the mean time you answered to me, so do you thing the power supply and the heat sinks are smallish?

Best regards

Eddie

[anatech]

Hi Eddie,
Without your amp on the bench in front of me, I can't say. In general all consumer amps have undersized heatsinks and power supply components. Some are much better than others. Yours is well build it sounds like, but you are bringing the operating parameters outside of where the designers set it up. So, measure the temperature differences between the two operating points.

I find that if I change something, I should leave it in the living room. If the change is audible, my wife and or kids will tell me sooner or later. It's the only true test. If there is no difference, choose the cooler operating point.

-Chris

[ilimzn]

Since the idle current in question is not high, and fairly small compared to all other currents drawn at rest, the impact of increasing it say, double, will actually be negligible, regarding cap and transformer life.

The reason why bias currents are often set with preheat is actually quite simple. Even though the bias current often has a small tempco, the reason why it can overshoot is because of the thermal delay between output transistors and bias transistor. The output and bias transistors are fairly small masses compared to the ehatsink. In thermal engineering, you have to figure each lump of metal divided by a boundary of different thermal resistance as a sort of thermal 'capacitor'. In a typical situation, you have at least 5 of them - power transistoir chip, power transistor tab, heatsink, bias transistor tab, bias transistor chip. You may also have number 6, enclosure trapped air (or similar airflow phenomena) around the heatsink.
In order to get the whole system to a stable point, preheat is performed. As Anatech said, DO NOT use 20kHz!!! Also Do NOT use a single sinewave and a speaker load, especially not over some few hunderd Hz as speakers are not designed to produce full power at a single frequency.

Regarding what you can expect with bias changes and sound, it's really anyone's guess. One could be a BIT more precise if one could see the actual amp topology, as some exhibit definite minimums in distortion related to bias currents - others do not. It is important to say that there is NO GENERAL CASE, it is simply false to assume that increasing the bias current will automatically produce better sound (even disregarding that 'better' is a wildly subjective measure). That being said, it CAN result in varius considerations and problems for the amp in question - maybe irrelevant, maybe potentially fatal, impossible to say without knowing the actual case in questiion.

[anatech]

Hi ilimzn,
That was well put.

I am more concerned with heatrise within the case. The ambient temperatures rise and have an impact on everything else.

-Chris

[eradoncic]

Originally posted by ilimzn
[B]

The reason why bias currents are often set with preheat is actually quite simple. Even though the bias current often has a small tempco, the reason why it can overshoot is because of the thermal delay between output transistors and bias transistor.

Sounds logical to me, and quite possible in this case. The interesting thing, initial measured value was fairly OK, around 7-8mV across 0R22, for both channels ant between both hot/cold amp (it is balanced push-pull amp)

In order to get the whole system to a stable point, preheat is performed. As Anatech said, DO NOT use 20kHz!!!

OK; we agreed with this, but I'm very curious why they are preheating with whoping 20kHz at around 60W, and for at least 5 minutes: I'm curius how much the heat sinks would be actually hot at that point of time (no, I'm not gooing to check that), since it is obvious that the bias point should be set at fairly abnormal temperature, at which the amp will rarelly go again during casual listening (and not party )

Regarding what you can expect with bias changes and sound, it's really anyone's guess. One could be a BIT more precise if one could see the actual amp topology, as some exhibit definite minimums in distortion related to bias currents - others do not.

You're right. The reason why I opened the amp which is otherwise in as new condition despite its age (according to serial number, produced in 1992), was too bold sound with emphesized bass lineage, resembling pure class B amps. I don0t know, but the amp was obviously not used, or used rarely, so maybe the capacitors were in question. However, adjusting a bit higher bias seems beneficial.