Some more Tripath Experiences

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
For those contemplating the EVM modules there is a lot of positive press - here's a little other side of the story.

I have built up a TA 0104A EVM module into a working amplifier. I've been reasonably careful and I've published my listening experiences earlier in this forum.

Here are some further experiences.

Ouput capacitor failure - like others my pwm filter caps failed promptly during test. It turns out they are rated 100V and are nowhere near up to standing the sorts of voltages the output filter can resonate to without a speaker load. Ok, the amp survived, new filter caps and it was ok I thought. On closer inspection however it had (naturally) killed the output filter damping resistors too (33R 2W). So, as an engineer I'm a bit annoyed that things as fundamental as this would get into the market - but I guess after all it is an evaluation module.

Radio interference is awful from the amp - both am and fm frequencies. At am frequencies it can be no other - there is about 3Vrms (130dBu) of switching frequency on the speaker lines after the filter. The CISPR standard for this on domestic power lines is 66dBu or 2mV! Spot the problem? Those of you concerned about noise getting into your delicate audio system be warned!! To be honest I had a Tact RCS2.0 room eq unit for a while and it certainly created rf interference as well.

Getting such gross levels of rf down to acceptable levels is a hard task for an experienced interference engineer. So don't expect to achieve miracles. I have included a mains rf filter, properly installed, recommended filtering on the output terminals, and added common mode filters to the line in - has a similar effect to transformers and is very effective at stopping rf heading out this way (by measurement).

Yesterday while trying to determine the greatest source of interference I ran the amp without speaker leads or speakers. At power down it killed two ouput mosfets blowing the 80V 10A supply fuses. This occurred as the 5V supply collapsed before the main supply - an issue Tripath's documentation mentions but says won't be problem. Well up to now it wasn't. The output circuit seems fairly resilient - I have replaced the mosfets and checked it over and it goes again. I have now added a small relay to force the unit into mute before the 5V supply decays.

I've seen mention of the amp as having great bass control - well with 801's you find out what has and what hasn't. It isn't bad but not up with top power amps. Mids and tops have a distinct character and lack of detail compared to a top amplifier (I’m comparing to a Plinius SA250/4 which I had hoped (optimist) that the tripath would oust - err .... wrong!).

On the other hand listening to it is pleasant enough - it doesn't hash or harder up treble like some amps. Perhaps for the $ it is a reasonable step but I suspect better value is to be had in the second hand market, without the problems or risks.

Just my experience - happy to answer questions. I am using it in my bass - it may not be as good as the Plinius but with it I can include a separate bass eq in circuit to notch out room modes which have hugely greater negative effects than the difference between the amps.

MurrayP, I've never seen a reference to the fact that the 5V supply needs to decay before the main rail supplies.
I think I read that if the amp is not muted and this happens, there could be some (nasty) transients into the speakers, but nothing like frying the output devices.

This could be one of the causes of my failures, because in my setup, due to the huge amount of filter capacity on the main rails they sure decay a long time after the 5V!

Can you retionalize on what happens and how to avoid it? I'd like to build an industrial strength amp, not one that will blow up at any moment!
I'm also planning on putting a relay on the output for further protecting my speakers. I guess this will increase the instability, right?

Last question: where did you get the replacements MosFets? I can't find them anywhere.
Sorry for my idear . I have just checked the Tripath TA0104 evaluation module (Regarding blowing output filter cap).

If the Cap is 1 220nF and the coil is a 18uH or if the cap is 100nF this should not get this amp up and smoking, but the design needs some cooling .. Not just 20degress/watt heatsinks..

Worst case switch current for 220nF and 18uH and +/-100V supply voltage is around 5.26Amps peak.

A lot of current to handle for an output filtercap .. Needs to be highquality like Philips,Wima, Rifa or ??

That is with no load and the switch frequency down at 200KHz
which the tripath is able to turn down at.

So the MOSFET will generate som heat. This and a tight deadband will generate more heat than the MOSFET are able to handle without a heatsink.

Maybe your amp is oscillating to .. Have to be checked with a scope.

Under all circumstances .. When one of your mosfets shorts it will put the rail voltage onto the outputfilter cap.

Murray, thanks for the post. I've been watching this amp design with interest, particularly, as you say, with the good press it's been receiving.

Notwithstanding the efficiency difference, how would you compare the Tripath with a good class A such as the Pass amps, or even a well made AB amp?
To avoid this problem again I have added an extra diode to the 5V supply bridge to block off the filter capacitor and placed a relay controlling the mute circuit directly across the bridge output before the diode. With the capacitor blocked by the diode, the relay immediately opens the mute circuit when the mains ac supply is removed. This is pretty simple to do. Incidentally when I was firing the amp up after repair I had 33R in place of the dc fuses - that worked fine until I powered it down without the mute on and it promptly vaporised the negative resistor. After that repeat warning I added the relay!

I used IRFP240's that I had lying around home to replace the ST devices. But the ST devices are readily available from both Farnell and Radio Spares.

The pwm filter caps blew at a much earlier stage than this recent failure. They blow because the output filter is not well damped without a speaker load and they can be severely over driven by resonance. I also doubt they are strictly rated for the ripple current as is suggested above. I replaced them with X2 250Vac PP filter caps - should really be foil types but the mains filter X2 types are much more readily available and didn't run too hot.

Running open circuit is a risky time for these amps - the filter (and so the current in the output devices) needs damping. A speaker protection relay seems like a good idea but does open circuit the amp. I have one but I'm nervous about it.

They are incredibly efficient amps, but sound wise are no match at all for my Plinius SA250/4 (class A) - I think I have made the differences pretty clear in my previous posts. At least in my implementation ther is no competiton and the rf issues are a complete pain to deal with. Others (eg John Curl) have also said as much.
Can you explain, please?

MurrayP, I'm very interested in the diode and relay mod you did. I'm not sure I understand how you did it, though.
Can you expand, please?

If I get it correctly, the Tripath, although good/warm sounding still is way below some of the best amps around (my MBL sounds a lot like your Plinius, just slightly "warmer").

Is there anything that I could build that sounds better than the Tripath?
Oh, I almost forgot, it should be powered out of a +/-75V supply :D because I've invested an outrageous amount of money on it and I don't want to throw it away!
Relay clarification (I hope!).

I added a relay which is effectively powered on when mains is present, and off (without delay) when the mains is removed. A normally open contact on the relay is placed in series with the mute circuit, so the amp mutes immediately when the power is removed (in particular it mutes before the 5V power supply decays so eliminates any risk of abnormal output mosfet switching activity at power down).

The relay has a 12Vdc coil. I power it from the bridge rectified supply of about 10V that feeds my 5V regulator. If I just connected it across the rectifer output the 10V supply smoothing cap would hold the relay in while the 10V and hence 5V supplies decay.

By adding a single diode after the rectifier and before the smoothing cap, you can isolate the smoothing cap from the bridge output and the voltage to the relay will immediately disappear on power down, releasing the relay. Works well.

Sorry don't have experience to help you with other amp recommendations.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.