Autotek KMX2000

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Autotek MX5000

Good morning everybody

I'm from South Africa, we have and amp named Starsound and model number SSA-D5000, its basically dual channel class D(X) version of the Autotek MX2000 from Perry's ART. Basically its a Autotek MX5000, as per pics vailable on the internet, they are identical in every way.

The amp came in with blown power supply and output transistors. After removing all fets and checking the drive waves in power and output sections, I noticed the power supply was fine so i fitted some test fets and continued with output.

when checking the class D switch, i found ther to be a coroded and severed VCC trace to one of the driver boards, this was the (B) channel and i ran a jumper from the driverboard pin directly to the high rail (which is where i calculate it to run to)

the one channel (A) wave forms seems perfect. when remote signal is applied and rails build, the class d switch begins after a small time delay on both high and low sides and rides along the rail voltage on the mosfet gates, the Vac RMS of the switching wave is 6.89v

the other channel (B) is a little different. when remote is applied, the rails build and the switch wave attempts to build alongside (A) but then the switching stops, after this the amp starts switching but the switch wave sort of ramps itself up over a time period of about 2~3mins and only reaches a maximum Vac RMS of about 5.42v on both the high and low sides. the shape of the wave somewhat approaches channel(A) as well but it just isnt the same. due to equipment constraints, i can't really zoom in and disect the wave.

a huge(5W) 0.22ohm (derived from symetry) resistor on the high side was also burnt to shreds when the amp came in and also replaced

i havent fitted any FETS in the output section, im used to seeing symetrical channels do identical "things", would imagine my channel (A) is ideal and channel (B) faulty. the (B) channel VCC trace was corroded and repaired too.

if anybody could offer advice/assistance on how to make my channel (B) work like channel (A), it would be greatly appreciated, i dont even know where to start.

Thanks in advance
Last edited:
Hi Perry

thanks for your reply. yes i have read thru this thread, however i will read thru it again just to be sure.

i removed both driver cards, most of the pins where supply voltage related and do receive their voltages.

4 pins where un accounted for and dont have any relation between the indiviudual cards. 2 of the pins where traced to the gates of the high and low side mosfets so i wouldnt suspect any trouble with those, the 3rd pins (symetrical pin on both cards) was input, and a clean sine wave appeared when signal was driven into the amp.

the last symetrical pin that was unaccounted for was the labeled "FB", which i assumed was feedback.

assuming both driver cards are working correctly, the functionality of this pin should be where my problem lies.

i will investigate further later on today and report findings.
hi guys

so basically this amp is, from research, built like a tank.

the only parts that failed were the PS fets and output mosfets.

the problem was, either the conformal coating is really cheap OR the glue is very acidic.

i read most threads relating to this amp, and one of them did mention to take out the glue and inspect the traces, i ignored this advice in an effort to try and maintain the "originality" of the amp. my aim with all my repairs is to achieve zero footprint while maintaining a solid repair.

given that the corroded trace was just at the edge of the glue, and after removing and installing the driver board 2ice and lifting a few traces later, i decided to rip the glue off and low and behold, 3 coroded traces.

after seeing 3 failed traces under just one mound of glue, i decided to rip the glue off the entire circuit and to my amazement, there were atleast 5 other potential points of failure ie corroding traces.

all traces have been repaired and both channels do identicle "things" so im fairly confident to fit fets in.

im going to make some PTFE/teflon base pads for the inductor coils so they are slightly elevated off the circuit board and dont vibrate/rub any traces short.

i will also paint around the area before i apply any adhesive.

i am still in a bit of a pickle tho, i dont know of any decent conformal coating to cover up my repaired traces just yet but i will find some soon and i prefer the glue used in these amps, i find it important that its sticks well when cool and removes fairly effortlessly when heated.

i also see a similar glue on speaker cones to secure the solder joint at which the VC wire and pigtails meet. its sort of a rubber texture glue. i dont even know what its called.

your advice on the glue and conformal coating would be greatly appreciated.

here in sunny SA, the names are definately different to what people may suggest over the internet, so maybe a detailed discription would be better so that i can match glue/conformal up at my local hardware/electronics stores.
You don't really need any coating, If you solder the trace to repair it, the flux will protect it. If you clean the flux, leave a small amount if you're concerned. If you want a colored coating, paint pens or sharpie markers will work.

For glue/fixative, glues like GOOP, E6000, ShoeGoo... will work. If you use silicone, use the type that's low odor (no acetic acid).

That beige fixative, when it turns dark becomes both conductive and corrosive. It should always be removed. I don't know whey they still use it. It's been causing ussues back at least as far as the early 1980's.

No need to use teflon or anything exotic. Plain old cardboard will work.

In the future, post images of the amp you're working on (not images from online). The fixative would likely have been mentioned in the first reply.
Hi Perry

thanks for the advice.

i will take a snaps and post soon, just so this thread is nice and complete to any potential repair techs with the same trouble.

i have repaired the traces and just left them bare.

as for my fixative solution, i went to my HW store yesterday and the only flex dry adhesive they had was some "acrylic paintable white sealant". i got myself a tube but i want to test it on some trial circut boards,some plain mild steel, maybe some copper as well to get a general feel of cohesion, adhesion, strength and acidity before i apply this to my amp.

this amp is a pure enthusiast rebuild, i bought it cheap in its blown state, and for now its going to be framed on my "wall of power" lol.

from the reviews, i hear this amp is a keeper, and its old as the hills so its not likely leaving my posession any time soon
Hi Perry

still with this amp, I'm chasing down a fault ,

At turn on the amp draws max current (~5amps via my current limited supply) until rails are charged and then idles at 1.8amps with one output fet/bank and every other component fitted.

at idle with one fet per bank and out of the heatsink, all the output fets get just warm to the touch, when i fit and clamp to the heatsink, the fets get less warm, i assume beause the heatsink is sinking the warmth.

when i turn off the amp, the high side fets receive a burst of warmth, as if a surge of current is being dumped thru them.

the only visible difference i can observe is that, on the mosfet gates, the speed of rail to rail oscillation is around 1.6khz on high and low sides of one channel But around 3.4khz on the other channel. comming back to my "identical channels doing identical things" i assume theres a problem here.

the previously damaged channel is the one currently oscillating at 3.4khz. but it seems that the 1.6khz high side heats up with "the surge" as well.

i have tried looking for "bumps" in the drive wave to identify if fets where being turned on fully at turn on, idle and turn off, but because i can only observe the wave in ac mode on my scope (wave rides on the rail voltages), there are many "bumps" and i cant really tell if its my scope cap normalising the DC component or the actual driver circuit sending a pulse.

if anybody could help, if of any relevance, which is the correct oscillation frequency of this amp output section and where does this oscillation start? i assume at the lm319, but i just cant be sure for now.

also any advice on my fets surging/ heating would be greatly appreciated.

im going scavenge in Perrys ART now now. but you guys input is always appreciated.

thanks everybody
Last edited:
You can't have two different oscillating frequencies on high and low since they're directly connected. That makes the frequency that you're giving questionable.

I've never checked these amps for a slight temperature increase on the high-side when powering down the amp. It could be normal. If both halves are doing it, I'd suspect it to be normal.
Hi Perry

I appreciate your reply and yes with the fets clamped to the heatsink, it does seem kind of negliable.

with regards to the switching frequency, i dont think i have explained adequately.

i have attached a few pics, this amp has 2 identical but isolated class D channels. the switching frequency on both high and low side gates of channel A are the same ie ~1600khz. the switching frequency on both high and low side gates on channel B is roughly double ie ~3600khz.

i decided to observe the R2R oscillation from remote turn on last night and it seems pretty instant that the channel A reaches ~1600khz while channel B sort of ramps up over ~45secs.

i have also read that similar switching frequecnies can cause undesirable noise so maybe this built in by design.

with your confirmation, if the frequency differences are normal as well, im going to source fets and fit them today and jump to a bench and endurance test later today.
hi Perry

i figured my cheapie scope isnt capable of measuring 1600khz. so i have double checked my scope with my signal generator,

channel A switch frequency is ~1.650khz(1650hz)


channel B switch frequency is ~3.450khz(3450hz)

is there a possibility that this is correct or should i hunt down a blown semiconductor?

the resistor/capacitor pair that should be responsible for creating the switching are exactly the same across the driver boards, the only possible scenario i can think of is there is a leaky transistor along the capacitor charging circuit that is leaking higher voltage than prescribed and therefore the cap is charging faster resulting in a higher switch frequency

or theres also that i could be chasing my own tail

im hoping you can judge for me
Last edited:
Perry you are really on another level!!!

not only did I have one but as it turns out, a million times more reliable than my cheap scope!!!

i knew this was going to bite me in the rear some day but i didn't think it was today!

none the less, i cracked it out and what do you know

Channel a switch freq ~98.4khz

Channel b switch freq ~96.8khz

i let the amp idle with just one fet/per bank all afternoon without signal, and its still idling with no excessive heat build up.

do you think im ready for full banks of fets Perry?
i must admit, i always knew my cheap scope was unreliable but i always assumed it would be out of tolerance by the same factor for any given measurement.

little did i think that in a space of 2khz i would get such ridiculously varying measurements.

needless to say, DMM is permanently the bench going forward.

im really great full Perry
Hi Perry

just an update on this amp.

with one FET/bank in the output section and everything else fitted, the amp turns on and initially draws ~5amps (max via current limited supply) for a second or so then the current draw falls to a healthy ~1.8amps..

just to be sure, i fitted and clamped everything in this state to the heatsink and without signal, let it idle for an entire afternoon. many hours later, all measurements checked out and having no excessive heat build up, i fitted full banks of output fets.

the amp powers up and draws ~5amps (max via current limiter) for a second or so and later settles around a healthy ~2.4amps at idle.

the power supply switching frequency is ~34khz
Channel A switching frequency is ~96khz
Channel B switching frequency is ~98khz

without an audio signal,the output to a high efficiency mid range driver has an extremely low volume high pitched whine. this whine is independent of gain level adjustment.

i have read in your ART that this is sometimes the case with class D testing using a high efficiency driver.

when fed with audio signal, the output signal is relatively clean (compared to a past HIP4080 repair). connected to a sub woofer, the whine is virtually indistinguishable even with the woofer right next to ear. playing a 10hz sine, the audio is really clean/satisfactory at low levels.

later today i will crack out the monster PSU and bench/endurance test to full power. i expect we have a successful repair.

thanks for your help Perry
Last edited:
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.