Secret shenanigans with an AU999

kevzep

Its all about the Music
Very busy at my real job so less time for this work, but a quick post for this Constant Current Source mod for the differential pair on the 999 power-amp.
I was trying to find a drop in solution, and didn't have much luck, but as luck would have it, I have hundreds of these LM334 which is used as a temperature sensor in the Crown Macro-Tech amplifiers. These devices can also be used as a Constant Current Source.
So I got to and came up with an adjustable solution on a perf board, put some legs on it, and slipped it onto the driver board, done.
I have only got as far as setting Bias and DC offset, I haven't heard it yet, but I hope to have a listen tomorrow night.
Heres a few pictures of the development. I'll explain a bit more about it when I have some time...
IMG_6676.JPG IMG_6677.JPG IMG_6680.JPG IMG_6681.JPG IMG_6683.JPG
 
That's nice, just 2 components = 1 current source.

I'm keen to hear how it performs.
 
Before I head into battle today, I am going to post a shot of the front end of the power amp to show the area we are working with, you can see how simple the design is, the supply rail just goes through that resistor R845, then through R811 and R813 into the Emitters of the Diff Pair and thats it, its coming from the 42V rail and runs through TR827 on F1231, this transistor is just for smoothing.
So when the amp pulls current, the rail drops, and we have an inconsistent current supply, especially with how dynamic music is, the current will be all over the place.
In a Differential pair, we need constant current, and in some of the later more high end Sansui's we can see them doing this by using separate supplies and Zener's etc etc to stabilise the supply.
Transistors are current devices, and in this application here as a differential pair, it couldn't be more important to have constant and equal current between TR801 and 803.
The idea is to have the current available for both transistors, with R845 not being a constant current source, the current is not going to be equal in the Diff pair because as the resistor 845 limits current and the supply is sagging in a transient, the bulk will go where its needed IE; TR801 and rob it from TR803, at least this is best that I can figure. This is not ideal and we can fix this.
Some of the results of inconsistent supply would be IM distortion, loss of dynamic range, phase shift, top end clarity and linearity...
If I get time later I will come back and explain more about how to go about this mod.
 

Attachments

  • AU999 Front end.pdf
    457.7 KB · Views: 160
Okay, so you take this LM334, mount it on the perf board. This is a TO92 case style.
Then you reference the data sheet to see the calc for working out your resistor. Its pretty much V=IR,simple as that.

You would have already measured the voltage drop over R845, and then worked out the current which is pretty close to 2.4mA.

Now you don't have to do this, but I chose to use a trimmer to set this constant current, just soI could get it precisely at 2.4mA. The trimmer I used is a 100Ω 25 turn.
Then I powered it up on my bench DC supply with my bench DMM in line to measure mA and dialled in the trimmer to 2.4mA.

I put some solid legs on the perf board to mount it in place of R845, you need to spread the legs a bit, or you could get more creative than me and get the leg spacing better, who knows I might revisit my solution although I think its OK. The V+ side goes to the power supply B5, then V- goes to the 56Ω resistor side.

Next step is to remove R811 and R813, these are only 10% tolerance, so they are not actually close enough in resistance, so out they go, and on you go with two matched 56Ω resistors, I just grabbed a pile of them and started measuring, they don't have to be exactly 56Ω but as long as they are a perfect match, don't mount them hard against the board, because later you need to be able to clip onto these to measure voltage drop.

I put a new matched pair of diff transistors in there, the ones that were in this were not that closely matched.

Once you have done this, then you can power it up.

Measure R811 and then R813 Voltage drop, they should each be about 65-66mV. The idea here is to get the voltage drop as close as possible on R811 and R813, I was able to get them within 0.1mV.
You do this by adjusting R807, it is a 3.9K, so on the two 999's I have done, I have had less current through R811 than R813.
So if you need more current on R811 you go down in R value on R807. It doesn't take much, and one affects the other, so when you increase current in R811, it decreases current in R813.
So the process is drop down one value, I went down to 3.3K, and then added resistors until I got the current equal.
In my instance it was a 3.3K, 430R, 100R. This got mine within 0.1mA. You could add it all up and use one resistor, but for the sake of accuracy, I just left it, I made kind of a hoop with the resistors. The 3.3 and the 430R are vertical mounted in the through holes, and the 100R bridges them. It actually looks fine, if you're good with your soldering its totally fine.
I was thinking perhaps of using a trimmer in parallel with the 3.9K resistor, but thats for later on.

And thats pretty much it.
I still haven't listened to it yet, my real job is in the way at the moment......maybe tonight, but I am just about to do a load test before I head to the office...

I will draw up a schematic at some point.
 
Last edited:
I had a listen to the 999 before I came into the office.....What is instantly noticeable, is the absolute extended range in the top end and the absolute linearity, its like the top end is brought to you on velvet, every detail is there and yet is presentation is so very smooth, its hard to describe...
There are little intricacies in the music that become more obvious, the imaging is insane.
Its next level guys!!
I can't wait to get home to do some more listening.....

This is a must for any AU999 owner.
 
Nice one Kev. This is exciting.

Any chance of some more pics, and a scribbled circuit diagram of the new transistor in-cicruit?

My reply to this post has gone awal since the site repair/maintenance.

A schematic will be drawn when I have time, but seriously, its one resistor which gets removed and replaced by the current source.
The instructions are clear and you can easily follow along by using the AU999 schematic, and by downloading the data sheet for the LM334, it has the specs and diagram for setting it up.

Its not a transistor we are using, its a Current Source, it has an input and an output V+ and V- respectively.

But if you need a diagram, I will have something done in a few days...
 
I just broke out my Kevzep modded AU-999 to put in my main system; sounds like I’ve got a bit of work to do!
 
Hmmmm, interesting, I thought there would be a lot more interest in this mod, its possibly the best one....
I think people are really interested Kev, I certainly am, but I don't own an AU-999. With the information provided so far, some brave soul should be able to pull this off, (it doesn't sound difficult to me), and give their impressions of the results. ;)

Either that, or you have blinded all the modification capable AU-999 owners with science. :) - or your previous mods are so good nobody want's to mess with something that sounds so good already. ;)
 
Last edited:
I think people are really interested Kev, I certainly am, but I don't own an AU-999. With the information provided so far, some brave soul should be able to pull this off, (it doesn't sound difficult to me), and give their impressions of the results. ;)

Either that, or you have blinded all the modification capable AU-999 owners with science. :) - or your previous mods are so good nobody want's to mess with something that sounds so good already. ;)
You can actually do this mod to any differential circuit, not limited by any means to the AU999, it can be done to 8080, 9090, AU888, even the Pioneer SX receivers like the SX1250 for example. It can be done to the AU9500.....
I might do another post including the models which it can be done to, this might expand the interest...

Plus I think you might be right, people are already satisfied....
 
Back
Top Bottom