Emotiva XPA-1 vs Jungson 99D

[AvMania]

Thanks for the pictures, the construction of the socket came very clear. The thing that you have heard acout the conductors is maybe because of the skin effect. Same rules apply to both line level and speaker level conductors though. Maybe I will use 20-22ga solid core conductors.

All those Mundorfs are humongous . Very sexy caps, I bet they also sound fantastic.

Hi Legis,

would you be able to help on the hum???

 

[Legis]

Hi AvMania, I' not sure what might cause the humming. It could be some bad joint or maybe some wiring issue. Power wires could go too close to big mundorf Mlytics and the 50/60Hz and it's harmonics jump inductively to them. Or then the humongous Mundorf signal caps are too close to chassis, or some other components and the hum jumps from there. Does the hum stay the same with XLR plug inserted and removed? Or then you might have connected the ground wire of the Mlytics to some other place than with you own XPA-1's, grounding is a mystery and you just have to try what spot hums the smallest. Propably somewhere close to the black star ground wires (and the ground point/center tap itself) could be a good place?


----------------
I have nowdone all the internal mods I intend to do, at least I like to think so . I changed the XLR sockets to same Cardas that AvMania also used (visible in picture), they are really fine sockets with a good locking system. As signal conductor I used teflon insulated UP-OCC from Neotech.

I also upped the bias current from 1,6A to 1,76A, that increased the Class A area from 40w to 50w (with 8R load) .

I performed the measurements with all mods for the first time couple of days ago. My dummy load exhibit some inductance, and I use 10kohm resistor to pad the input signal before the sound card (that generates THD and noise at some extent), so the THD at high freqs can be somewhat higher than with optimal equipment. Also the EMU 0404 USB's DA and AD circuits itself distort more than in real pro equipment.

The voltage readings are way off, but they still act linear, meaning if you reduce the voltage on the picture to half from the high value (that is 21V), the actual voltage and in real life was half of this, ie. 10,5V.

An externally hosted image should be here but it was not working when we last tested it.

Distortion versus output voltage with 374Hz sine (0,05V-21V / from milliwatts to 110W@4R / 55W@8R). 8R is lower, 4R is higher.

An externally hosted image should be here but it was not working when we last tested it.

Distortion versus output voltage with 20kHz sine (0,05V-21V / from milliwatts to 110W@4R / 55W@8R). 8R fares marginally better.

An externally hosted image should be here but it was not working when we last tested it.

Harmonic structure at 8R, 10V (12,5W):

An externally hosted image should be here but it was not working when we last tested it.

Harmonic structure at 4R, 10V (25W):

An externally hosted image should be here but it was not working when we last tested it.

- As can be seen from the first THD picture, the XPA-1 now pushes all the way to the rare 0,000x distortion reqion, where even pre amps rarely go. The THD and bass and mid freqs drops to 0,00045% between 0,8-3,2W of output power at 8R load and 0,0007% at 4R load. As stock the THD minimum was approx. 0,001% at 8R and 0,002% at 4R. Increasing the bias "from 3W to 50W" reduces the THD across the power spectrum, not with just 3-50W power levels (what some might usually think).
-Harmonic structure is based on low harmonics (H2 and H3) and upper harmonics are suppressed quickly. Increasing bias dropped higher harmonic content quite much.
- IMD also dropped (I have no pictures) across the scale, which shows that IMD usually correlates with THD.
- Added capacitance helps filtering the 50Hz and it's harmonics, but the increased bias and operating temperature generates naturally some "heat noise", so the net effect regarding S/N ration is propably zero.

 

[AvMania]

Hi Legis,

Yes when the xlr is inserted it increases.

 

[AvMania]

You menision the signal cable. Which are they? White cables running to each board. I would like do this as well.

I managed to get rid of most of the hum. now there is a low hiss which gets worse when the xlr is plugged in. Also the signal is much softer than the other amp. Does this give any clue?

Thanks for your help so far.

 

[Legis]

You menision the signal cable. Which are they? White cables running to each board. I would like do this as well.

I managed to get rid of most of the hum. now there is a low hiss which gets worse when the xlr is plugged in. Also the signal is much softer than the other amp. Does this give any clue?

Thanks for your help so far.

Hi, by signal condutor I mean the ~1,5" long wires that connect to the XLR socket . White cables deliver AC to each amp module for the microprosessor circuity.

By softer signal do you mean not as loud or does it relate to sound quality? I would triple check all the joints, especially on the PSU/input card and signal caps. Check the continuity with multimeter if unsure. How much louder is the hiss than normally? How did you manage to get rid of the hum and did it, whatever you did, cause the hiss or was it there all the way long?

 

[AvMania]

Hi Legis,

Amp A is dead silent. No buzz or hiss when I stand next to it and put my ear next to the tweeter or midrange.

Amp B had a loud Hum and the volume of the hum did not increase or get less when the volume on the preamp is changed. Also when I tested each channel, Amp B was softer, the gain was lower.
I was at a loss as I had done 3 of the others the same way and no problem. So I thought if may be the power board and the Red and Blue wires running underneath being now close to it as opposed to previously being shielded by the metal plate, now removed. I connected an old speaker and on my workbench had it open powered up (I know..) and was pushing in a non conductive foam (hard type) to separate the two when the hum went. This got me to try every wire and I found one of the two black (ground) cables was the culprit. I managed to fix this by adjusting the clip and it seems to work but as the hum goes a lower softer hiss becomes apparent. I have not had any luck with the hiss. I have removed the caps (4.7’s) and its still there. fixed it all back and connecting it back to the pre and the hiss came though the main speakers and the gain was still low.

I also found that when I touched the caps a bussing sound appeared. Same with the touching the body of the XLR socket when it’s plugged in. I have not tried this on the other amp to see if its normal.
Does this give you any idea what is going on?

Also the way I connected the XLR’s was to solder a piece of bare 1mm silver wire from the board to the XLR. It’s quite stiff and holds it in place even when the back plate is removed.


Thanks again

 

[AvMania]

can you tell me how the signal goes from the main board to the negative and postive amplification boards on each side. surely its not on the ribbon cable?

 

[Legis]

can you tell me how the signal goes from the main board to the negative and postive amplification boards on each side. surely its not on the ribbon cable?

Yep, the signal goes in the ribbon cable to the amp modules. Makes changing them little harder .

Hopefully nothing has not been burned inside the unit. Can you test the amp modules/boards (left and right) individually with the scope you own, to see wether both modules output the signal normally? Just measure from speaker terminals while playing for ex. sine wave. You don't have to connect the alligator ground clip of the measurement probe anywhere, especially do not connect it to the other speaker terminal. It would explain the reduced gain, if one amp module is malfunctioning or is not getting signal. This would reduce the voltage swing to half, and effectively reduce the gain by 6dB.

 

[AvMania]

Yep, the signal goes in the ribbon cable to the amp modules. Makes changing them little harder .

Hopefully nothing has not been burned inside the unit. Can you test the amp modules/boards (left and right) individually with the scope you own, to see wether both modules output the signal normally? Just measure from speaker terminals while playing for ex. sine wave. You don't have to connect the alligator ground clip of the measurement probe anywhere, especially do not connect it to the other speaker terminal. It would explain the reduced gain, if one amp module is malfunctioning or is not getting signal. This would reduce the voltage swing to half, and effectively reduce the gain by 6dB.

Ok I will do it this afternoon after work (7 hrs time) . So is there only one connection? just the positive to one terminal at a time?

I'm wondering if I should makeup the signal cable. get a new one just in case? Would you remeber how many lines on the ribbon cable?

Thanks

 

[AvMania]

Hi Legis,

Yes when the xlr is inserted it increases.

Why would this be the case?

 

[Legis]

Ok I will do it this afternoon after work (7 hrs time) . So is there only one connection? just the positive to one terminal at a time?

I'm wondering if I should makeup the signal cable. get a new one just in case? Would you remeber how many lines on the ribbon cable?

Thanks

Yes, just the tip of the measurement probe to one terminal at the time and check the p-p or rms voltage of them both so you can determine the correct board and concentrate on it.

If you find out that other board does not output anything, then empty the capacitor banks for example with a light bulb modded with copper legs so that they are completely depleted and measure the resistance between all the power and driver transistor's legs. The DC has to be drained completely, some meters don't work correctly if there is charge left. Compare the values to the one board that is working.

I don't know why the hiss/hum increases after inserting a XLR plug.

According to my old pictures there are 10 printed lines on a ribbon cable's socket but I think that each line has two conductors, and therefore the ribbon cable might have 20 conductors.

 

[AvMania]

Hi Legis,

In a few posts back you posted some measurements, noting THD was at preamp level. What were the changed done between this test and the previous time you made a post?

In terms of listening what were the changes?

Thanks

 

[Legis]

Hi Legis,

In a few posts back you posted some measurements, noting THD was at preamp level. What were the changed done between this test and the previous time you made a post?

In terms of listening what were the changes?

Thanks

Hi AvMania,

because my sound card had been broken for a long time, the measurement basically regarded all the latest mods (bias current, signal caps, PSU caps, XLR-socket etc.). It was the first time I measured the amps with fully modded condition. But the increased bias current is definitely the one mod to blame for lower THD. It seems to drop harmonics 2 and 4 (H2 and H4) and also upper even order harmonics. It does not affect that much to the H3 and H5, because I kept the bias voltage the same, meaning that the voltage drop accross a emitter resistor is approximately 11mV. Well maybe it's a tad higher than originally (which was 10-10.5mV). Going from 0.5R resistors to 0.15R, and keeping the bias voltage around the original, increased the bias current from 0.48A from 1.76A, translating into an increment from 3.5W to 50W Class A region into 8R resistive load (1.76A x 2 x 4R) x 1.76A x 2, if I have calculated it correctly.

All the prior listening comments are still valid, some of them were given with 40w Class A, and some with 50W Class A region, but that's not a big deal. I really like what the increased bias does to the sound. Definitely the best mod for me, but also 1000x the hardest because the top cover has to be modded with better ventilation also. The cycle of measuring, monitoring, measuring is quite intense. I had to monitor the bias and temperature in different conditions and it took approx. 1 week to make sure that everything is right, temps keep themselves under control even in equipment rack and top cover on, biases are optimally set even after 12h after power up and so on. Should I say more...

By the way I measured the maximum unclipped RMS output voltage swing into 4R load with a scope. On a scope one can see only approx. 1% distortion. The signal burst was 800ms long and I tried with different sines between 20Hz and 20.000Hz. Outputted voltage and was approximately the same with every frequency so the XPA-1 has full power bandwith (at least) of 20-20.000hz, which is respectable feature.

The sine waves came out unclipped at approx. 66Vrms, which equals to ~1090W. I think this is very good number because my mains voltage drops quite much under load. 2R test should be very interesting, the only problem is that how to construct a dummy load that can handle such power and is still quite cheap.

 

[AvMania]

Hi Legis,

Thanks for your feedback. My amp is up and running again. stunning is all I can say.

 

[Legis]

I measured frequency response of my modded XPA-1s with new function generator that I got. The response of XPA-1 was attenuated 0.48dB at 100kHz, 1.96dB at 200kHz and 4.7dB at 300kHz:ssä. The -3dB is somewhere around 250kHz. The input's 10µH inductance filter kicks in almost at the same point and the response starts to fall more rapidly after that.

My XPA-1s are quite more faster than stock (according to measurements that others have done) because I have upped the bias (transistors operate faster) and changed the emitter resistors to low inductive Mills (emitter resistor's inductance might start to affect the response at those frequencies).

Here's 30kHz square wave looking very good

An externally hosted image should be here but it was not working when we last tested it.

And 100kHz Square wave, whick starts to look more of a like clipped sine wave because the fundamental frequency gets so near the 10µH inductance filter's xo-point.

An externally hosted image should be here but it was not working when we last tested it.

I could find couple of amps at stereophile that are less attenuated at 200kHz than modded XPA-1 (Soulution 710 and Marantz SM11-S Refence). Modded XPA-1s are actually quite fast.

I read from Pass's F5 turbo paper that balanced amps retain their speed better when paralleling output transistors. 24pcs of output transistors at the same channel would make quite slow amp, but when doing balanced 12pcs + 12pcs amp both halves are much faster, and the slew rates of the halves are actually added together.

By the way I used only one half of the XPA-1, because the function generator does only have single ended output. I plugged the signal to the XLR output and shorted the other phase of the input to the ground. XLR input was faster than the RCA input because RCA input has an inverting op-amp.

 

[AvMania]

Hi Legis,

Can you tell me how you made the XPA-1 operate at a higher power in CLass A mode?

thanks

 

[Legis]

Hi Legis,

Can you tell me how you made the XPA-1 operate at a higher power in CLass A mode?

thanks

I think have explained this in older posts, so check them out also.

One needs to change the emitter resistors to lower values (I used 0.15R) and keep the bias voltage the same, around 10mV. Just upping the bias voltage (and keeping the original 0.5R emitter resistors) increased crossover distortion, 9-11mV is optimal value for bias voltage. One needs to modify the case, the original metal chassis do not breath too well. If one wants to keep the chassis the same, then maybe 0.25R resistors and 10mV would be better to prevent overheating. 0.15R and ~10-10.5mV bias voltage gives approx 50W Class A, this is the setup I use.

 

[a.wayne]

Good work legis and thanks for the update ........

 

[Legis]

Reducing the output Zobel's (located behing other speaker terminal = the small resistor and capacitor on a miniature PCB) capacitor to 0,1µF increased the cutoff frequency of the filter so that the -3dB point of the amp increased from ~250khz to ~410khz. I use 10R Mills resistor and Evox Rifa's PHE 426 pulse capacitor, that has low self inductance. The original 0,22µF cap is unnecessarily high value and it's cutoff frequency limits the bandwidth of the amp (at least when the output transistor's speed is boosted with bigger bias).

 

[AvMania]

Reducing the output Zobel's (located behing other speaker terminal = the small resistor and capacitor on a miniature PCB) capacitor to 0,1µF increased the cutoff frequency of the filter so that the -3dB point of the amp increased from ~250khz to ~410khz. I use 10R Mills resistor and Evox Rifa's PHE 426 pulse capacitor, that has low self inductance. The original 0,22µF cap is unnecessarily high value and it's cutoff frequency limits the bandwidth of the amp (at least when the output transistor's speed is boosted with bigger bias).

Is there an audible difference once this was done?