Author Topic: SOLVED:Catastrophe on an Emotiva XPA-3 Gen 1 amp  (Read 3262 times)

0 Members and 1 Guest are viewing this topic.

Online dietert1

  • Super Contributor
  • ***
  • Posts: 2018
  • Country: br
    • CADT Homepage
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #25 on: January 31, 2023, 09:49:27 pm »
The P3000 circuit shows current limiting. Output current of each transistor is measured on 0.47 Ohm emitter resistor (R1 .. R6). Then R7 .. R12 get the average current of all six output transistors as voltage. Then there is a 4:1 divider (R188 R187). The the current limit is established using the base-emitter voltage of Q57 as a threshold. It becomes conductive at about 0.6 V. times 4 = 2.4 V, divide by 0.47 Ohm gives about 5 A per transistor or 30 A in total. When it becomes conductive, it clamps the input voltage of the output stage, near the bias circuit. This limiting circuit is present 4x in a stereo amp.
To get a current limit proportional to output voltage one can add a resistor from the base of Q57 to Gnd. For example delete R187 and replace it by a 2.2 KOhm resistor in series with a diode from the base of Q57 to Gnd. Then the current limit will be about 1.2 A per transistor near and below 0 V output voltage and 5 A per transistor near 60 V output. At 60 V output the output transistor has only 8 V collector-emitter voltage, so at 5 A it only takes 40 W.

Regards, Dieter
 

Online Audiorepair

  • Frequent Contributor
  • **
  • Posts: 697
  • Country: gb
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #26 on: January 31, 2023, 10:38:35 pm »
The R188/R187 divider is shunted by an NTC and fixed resistor, R200 and R189.

My simple maths calculate that as the temperature increases, the limit threshold is therefore increased, which seems somewhat counter intuitive.  (?)

(I presume the NTC is on or near the heatsink)



Edit:  Er, to clarify, I mean that with increasing temperature comes less limiting, not more.
« Last Edit: January 31, 2023, 10:44:53 pm by Audiorepair »
 

Online dietert1

  • Super Contributor
  • ***
  • Posts: 2018
  • Country: br
    • CADT Homepage
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #27 on: January 31, 2023, 10:59:32 pm »
Yes, very strange. Maybe they want to compensate the -2 mV/K temperature coefficient of the base-emitter threashold voltage of Q57.
 

Online Audiorepair

  • Frequent Contributor
  • **
  • Posts: 697
  • Country: gb
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #28 on: January 31, 2023, 11:01:56 pm »
Or NTC is a typo for PTC?

Or maybe its not actually measuring the heatsink then and it is just a tempco as you suggest.
« Last Edit: January 31, 2023, 11:15:35 pm by Audiorepair »
 

Offline JeremyC

  • Regular Contributor
  • *
  • Posts: 135
  • Country: us
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #29 on: February 01, 2023, 01:07:45 am »

Yes I did and both are alrite on a DMM check......have ordered the final stage BJT's from Newark and the other 4 from Ebay :palm: as they were available nowhere else and hope to change them out sometime within the next 2 weeks.

Most of the LCR meters (including DMMs) are testing the DUT under low voltage. I suggest you replace just in case these 2 caps with quality, 200V+ rated.

Emotiva has indeed incorporated a dc protect using a UPC1237.
Once I get hold of the damaged parts, I am thinking of putting in a current monitor based on the individual emitter resistor V drops  and OR them to shut down the PS relay......I believe there was something similar done in an Elektor magazine some years before.

Any other ideas on how to keep the output BJT's always under the SOA curve?
To their credit, Emotiva has built the amp extremely serviceable but getting hold of the o/p BJT's especially the NJW0281G has been a pain......only Newark seem to have them.
Aliexpress and Ebay seems to have tons of them but many are missing the "ON" semiconductor logo.    :scared:

Another suggestions:
- Don’t trust anything from Ebay, the vendors are famous for faked components , use Mouser, Digi-Key, Newark etc. They maybe little bit more expensive, but you will be sure what you are getting, especial when comes to paired transistors...
- Improve the Emotiva’s design and add separate fuse(s) for each channel’s rail. I would start with underestimated fuse like 5A. It maybe frustrating, but it will save your $$$.
- Try power on the amp after repair in series with 200W bulb (I mean classic bulb and not an LED) and ammeters in the rails.

Good luck and keep posting how it went.

« Last Edit: February 01, 2023, 01:41:24 am by JeremyC »
 

Offline JeremyC

  • Regular Contributor
  • *
  • Posts: 135
  • Country: us
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #30 on: February 01, 2023, 01:29:27 am »
Any other ideas on how to keep the output BJT's always under the SOA curve?

Contact Mouser, or Digi-Key, they offer paired transistors.

BTW, did you ask Emotiva, they may have solution for you.
« Last Edit: February 01, 2023, 01:31:28 am by JeremyC »
 

Offline rfenggTopic starter

  • Regular Contributor
  • *
  • Posts: 127
  • Country: us
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #31 on: February 02, 2023, 02:20:34 pm »
Any other ideas on how to keep the output BJT's always under the SOA curve?

Contact Mouser, or Digi-Key, they offer paired transistors.

BTW, did you ask Emotiva, they may have solution for you.

Yes I did contact Emotiva......they offered a flat $250 for the repair.......I was quite alright with that but the shipping is prohibitively expensive......the amp weighs close to 55lbs thanks to the conventional transformer based PS.
 

Offline rfenggTopic starter

  • Regular Contributor
  • *
  • Posts: 127
  • Country: us
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #32 on: February 02, 2023, 02:24:36 pm »

Yes I did and both are alrite on a DMM check......have ordered the final stage BJT's from Newark and the other 4 from Ebay :palm: as they were available nowhere else and hope to change them out sometime within the next 2 weeks.

Most of the LCR meters (including DMMs) are testing the DUT under low voltage. I suggest you replace just in case these 2 caps with quality, 200V+ rated.

Emotiva has indeed incorporated a dc protect using a UPC1237.
Once I get hold of the damaged parts, I am thinking of putting in a current monitor based on the individual emitter resistor V drops  and OR them to shut down the PS relay......I believe there was something similar done in an Elektor magazine some years before.

Any other ideas on how to keep the output BJT's always under the SOA curve?
To their credit, Emotiva has built the amp extremely serviceable but getting hold of the o/p BJT's especially the NJW0281G has been a pain......only Newark seem to have them.
Aliexpress and Ebay seems to have tons of them but many are missing the "ON" semiconductor logo.    :scared:

Another suggestions:
- Don’t trust anything from Ebay, the vendors are famous for faked components , use Mouser, Digi-Key, Newark etc. They maybe little bit more expensive, but you will be sure what you are getting, especial when comes to paired transistors...
- Improve the Emotiva’s design and add separate fuse(s) for each channel’s rail. I would start with underestimated fuse like 5A. It maybe frustrating, but it will save your $$$.
- Try power on the amp after repair in series with 200W bulb (I mean classic bulb and not an LED) and ammeters in the rails.

Good luck and keep posting how it went.

Thanks JeremyC......I found the final NJW's from Newark ($3 each) but the 2 of the  other 4 I need i.e., 2SC4793, KSA2073, 2SA1837, KSA940 are hard to come by from Mouser/Digikey/Arrow/Newark.
Will try to implement separate fuses as well.......I do use a dim bulb tester with an incandescent 100W in series when I do the repairs.
 

Offline rfenggTopic starter

  • Regular Contributor
  • *
  • Posts: 127
  • Country: us
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #33 on: March 11, 2023, 02:59:09 pm »
She finally sings again !!! :box:

After waiting for what seemed like an eternity, I managed to get hold of originals all the damaged parts and replacing them has made the channel 2 of the amp completely functional again; channel 1 and 3 were already working fine but the short on channel 2 kept blowing the 10A fuses......


A humble thanks to everyone on the forum  whose advice was invaluable.


Also replaced the caps as per @Tomorokoshi advice, just in case as they were cheap and easily available from Digikey.

Cheers to all  :-+
« Last Edit: March 11, 2023, 03:03:33 pm by rfengg »
 

Offline elecdonia

  • Frequent Contributor
  • **
  • Posts: 399
  • Country: us
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #34 on: July 12, 2023, 06:13:37 pm »
Maximum collector current isn't the criterion, but the SOA (safe operation area). In Figure 2 of the NJW0281 datasheet you can see that at 130 V it will stand about 2 A collector current for about 10 msec. This is the situation when you decelerate fully accelerated bass cone by reverting voltage from +65 to -65 V or vice versa. If the speaker has 4 Ohm resistance, you need about 32 A, so that gives 16 transistor pairs and they should be "cold" before you try. As the deceleration isn't a rectangular but a triangular power pulse it may survive with 8 pairs - when cold!
Regards, Dieter
SOA (safe operating area) is a critically important specification for analog audio power amplifiers. Unfortunately SOA isn't well understood. The semiconductor manufacturers tend to promote specifications like "maximum current" and "maximum voltage" because larger numbers look good. Therefore one must search through the transistor manufacturer's datasheet for the graph of "Safe Operating Area" to determine the (much lower) allowable collector current for large E-C voltages.

Nearly all of the BJT power transistors made for linear audio amplifiers have similar SOA characteristics. Allowing roughly 2A per output transistor is a good rule of thumb for all linear audio amplifier circuits with power supply rails of +/-50V and larger. This explains why well-designed amplifiers intended for "heavy duty" usage have such a large number of output transistors arranged in parallel.

Another circuit design known as "Class H" uses multiple +/- power supply rails along with commutating diodes and additional transistors which momentarily raise the +/- rail voltages for large peak audio signals. Carver and NAD were the first manufacturers to popularize this technique. Typical rail voltages are something like +/-30V and +/-60V. For small amplitude signals the main output transistors receive collector voltages of +/-30V. This is below the "2nd breakdown" portion of the SOA chart where the maximum permissible collector current drops rapidly. When a musical peak would be clipped with +/-30V rails, the Class H circuit briefly raises the rail voltages to +/-60V. However, because the instantaneous output voltage is already close to 30V, the actual E-C voltage across each main output transistor never exceeds 30V.

Some of the Carver amplifiers with huge power ratings have 3 sets of supply rails: +/-30V, +/-65V, and +/-120V. Again, the higher voltage rails are applied to the main output transistor collectors for only a few msec at a time. 
I’m learning to be a leading-edge designer of trailing-edge technology.
 
The following users thanked this post: rfengg

Offline elecdonia

  • Frequent Contributor
  • **
  • Posts: 399
  • Country: us
Re: Catastrophe on an Emotiva XPA-3 Gen 1 amp
« Reply #35 on: July 12, 2023, 06:46:15 pm »
Well this is how the big boys do it, might give you some ideas. Page 14 shows the power amp schematic.
This Electro_Voice P3000 2-channel power amplifier has an unusual output stage design where the main supply rails float relative to ground. The output configuration is a "full H bridge." However, unlike most bridged amplifier circuits, one H-bridge output is grounded while the other H-bridge output drives the load. This is why the power supply rails must "float." Each channel of this Electro-Voice P3000 amplifier has 24 main output transistors which are arranged in 4 groups of 6 parallel transistors each. 

The total supply voltage of 150V is divided by 4 for each of the 4 sets of output transistors. This is 37.5V for each of the 4 legs of the H bridge. 37.5V is below the "2nd breakdown" voltage of the output transistors.

I believe the oldest implementation of this design might have been known as the "circlotron."

A stereo amplifier of this design requires two main power transformers (or two isolated secondary windings on a single power transformer) because of this "floating" design. However, only a single bulk energy storage capacitor (or group of capacitors) is needed per channel. The main supply rails aren't +/-75V. Rather this is a single-ended 150V supply rail with no reference to ground.

A legendary American stereo power amplifier known as the Electron Kinetics Eagle 7 (or 7A) had a similar circuit topology. In this case there are 12 output transistors per channel arranged in 4 groups of 3 parallel transistors per group. 
« Last Edit: July 12, 2023, 06:58:26 pm by elecdonia »
I’m learning to be a leading-edge designer of trailing-edge technology.
 
The following users thanked this post: rfengg


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf