My JLH 1996 EVO....output transistors questions....

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
(Please don't kill me saying there is another JLH thread. I have been reading that thread day after day and my wife is about to divorce me and let me marry JLH! I learned a lot from that thread but it is out of control and has 227 pages !)

After my all MOSFET amp disaster, I focus on the JLH. I have just built the 1996 EVO based on Rudy's schematic. Simple and no CCS, keeping the boost strap etc... My goal is also to use as much parts as I can find from my parts bin. Only thing I bought are fake Toshiba output transistors because my 2n3055s are all different brands and not getting a match for all the fours I needed. I want to make sure it works before I put better transistors.

I put the prototype together one channel at a time using a pair of original motorola 3055. The trans have great gain and good sound. I ran the prototype off around +-13.5V DC (from a small 24v CT transformer). Then I replaced with the fake Toshiba. They have very little gain (around 10 hfe) so I have to adjust the current up until above 700mA or so before it started to sound good.

Today I moved the prototype into the enclosure with a bigger 16v CT tranny so I'm getting approx +-23v dual rails. I have the Iq set at around 1.3A now. I have to use active cooling to keep the output transistors temp in control. 0mV DC offset without drifting.

Just as I think I'm done and am listening to the amp, I just feel like the amp does not sound as warm as when I ran it with +-13V with a little lower Iq. (with same speakers and same source)

I know at this time you're probably about to say change out the fake toshiba (which is what I'm planning to do) to 15003. I still have to order some more thermal grease while I'm ordering some 15003.

But my question is - "Is current always the higher the better? Is supply voltage always the higher the better?"

Also, if I use 15003 in place of 3055, will the heat reduced due to the bigger power dissipation from a bigger size semiconductor chip?

This question is about paralleling output transistors: Say, if I have a single transistor and set its Iq, say at 1A to put it in pure class A. And say, the same transistor will be in class AB when its Iq is 0.5 A. Then I parallel two of them, keeping total current to 1A and each transistor carries 0.5A. Will the final product in class A or class AB ? The reason I'm asking this is I have read people paralleling transistors to reduce heat generation and increase output power. I'm wondering if letting two transistors sharing the current will make them running in a less than linear region compare to just use one transistor and set it at a higher current.

Thanks.
- AC
 
I built a 1996 JLH and it's awesome!
I used 15003s and +/-16.5V rails. Iq is between 1-1.5A (depending on how I've been tweaking it at the time).
I feel that any less than 1A and it doesn't sound as good.

If you're 'underbiasing' it, it doesn't force it into class AB. I think it sort of clips one cycle (and the DC offset adjustment would be tweaked to centre it around 0V). I'm sure someone can define more accurately what happens, but you definitely don't get crossover distortion like in class B/AB.

I think you've answered your own question, moving to MJ15003 should solve your issues. Mine sound fine at 1A (I haven't paralleled mine).

If the 3055 is a plastic package, and you're moving to a metal can MJ15003, then yes, the heat transfer should be better.
 
69JLH is class A only, will not do AB at any bias setting.
Current steering with BJTs defines class A with a hump
in the middle that is totally opposite the dip required for
AB operation. When driven to extreme it simply clips.
At no time or setting will it go AB. NEVER!

The hack to define an AB style output class requires a
non-linear pre-shaping of total available drive current
before JLH steering. Shunting away excess bias, beyond
that which defines your chosen output class, is probably
the easiest way. The choice of bootstrap or CCS to feed
such a bias shaping shunt is a matter of personal taste.
 
But my question is - "Is current always the higher the better? Is supply voltage always the higher the better?"

The main issue with higher current and/or higher voltage is power dissipation. Otherwise, my feeling is that higher voltage is usually better. I don't have enough practical experience to comment too much, but I would expect higher voltage to lead to lower distortion.

The thing about 2n3055 in terms of current is that you want to avoid the current being so high as to enter the region where these devices have excessive beta droop. All BJTs suffer from this to different extents. So parallel devices like you are planning to use will allow each device to operate at lower idle currents where the gain is still good.

I plan to stick with 2n3055s for mine and operate each device around 1/3A with 4 devices in parallel for a total of 4/3A.
 
Thanks guys for comments. I will definitely go put 15003 later and set around 1A as Phil suggested. And thanks Bigun for explaining the paralleling transistors. I didn't pay much attention to beta droop but now I understand why. My fake Toshiba has very low hfe (around 10 or 11, I tested them on breadboard). So when I raised Iq to 1.3A, it further dropped the beta and the sound became compressed. Ken, appreciate your input but I don't think I get it.

So, I lowered the Iq to 1A today. The sound is fantastic ! Even with fake Toshiba. Already better than my MF A1s. I think with the toshiba, anything higher than 1A will make the hfe too low to sound right. Anything below 700mA also make the amp sound muff and dry (even my wife can tell). I have been listening to it now for 3+ hours and not feeling tired. Another good side effect is the heat dissipation has reduced with less Iq. But still needs active cooling. I have some silent computer fan which I can hardly hear it running and it does the job well. The fan in the pic is an 120mm but I do have silent 80mm fan to install. As you can see from the pics, it is still not completely finished. The enclosure is too big I think. You can also see I use 2n4232 as the phase splitter. It doesn't have a high hfe (around 50) but works fine. The input PNP is a no name type that a relative passed onto me 35+ years ago (hfe around 170). The boost trap cap is phillips (blue) and the input cap is a 0.68uF (westcap, yellow). Bass reproduction is not a problem with this yellow cap. You can see I mounted the 4 toshiba to a L bracket then bolted it at the back with heatsink. Then I put a fan over the heatsink. It works well. Just not sure if heat will be a problem when I have the top panel install later. Power supply is a simple 36VCT6A transformer with bridge rectifier on two 6,800uF caps. There are just a little hum when I am very close to the speaker. C-R-C filter will be installed later.

Sound is great. Very very warm, especially the female vocal. Bass is very strong. Clarity is surperb in the entire frequency range. I'm only using my ipod as source with MP3s I ripped or internet streaming. It has already surpassed my expectation. I'm not even sure what will happen if I change the 3055s to 15003s (maybe making my hairs at my back standing straight ??).

- AC
 

Attachments

  • DSC01680.JPG
    DSC01680.JPG
    903.8 KB · Views: 1,246
  • DSC01681.JPG
    DSC01681.JPG
    880.3 KB · Views: 1,174
  • DSC01683.JPG
    DSC01683.JPG
    901.6 KB · Views: 1,126
  • DSC01684.JPG
    DSC01684.JPG
    880.9 KB · Views: 1,105
  • DSC01685.JPG
    DSC01685.JPG
    885.8 KB · Views: 1,054
Few more pics here......

I have done a little more on the amp mostly finishing up and here's a few more pics.

I mounted a front power sw with built in light (left) and a volume pot (right). Then I permanent mount a 80mm silent computer fan (got from newegg) powered by a cheap 12v wallwart mounted inside the enclosure. It keeps the output transistor at a good temp and will not overheat.

I found myself enjoying the amp everyday even just with the low hFE fake Toshiba. (Haven't receive my 15003 yet). My other remaining minor issues are trying to add a pair of filtering caps to form a CRC filter to get rid of the tiny hum from power rails and a slight vibration top panel due to the location and close proximity of the power transformer (links below).

http://www.diyaudio.com/forums/power-supplies/209549-i-thought-myth.html
http://www.diyaudio.com/forums/power-supplies/209550-chassis-vibration-how-solve.html

Rarely do I fall asleep while listening to music. But this JLH amp sounds so warm and soothing and it repeatly put me to dreams everytime I have it turn on !

-AC
 

Attachments

  • DSC01868.JPG
    DSC01868.JPG
    141.2 KB · Views: 349
  • DSC01869.JPG
    DSC01869.JPG
    142.6 KB · Views: 207
  • DSC01870.JPG
    DSC01870.JPG
    142.5 KB · Views: 199
  • DSC01872.JPG
    DSC01872.JPG
    137 KB · Views: 210
Initially I have two pairs of antique Motorola 2n3055 but I found one of them has leak between C-E. I went ahead and use the good pair to finish two channels (one at a time). I have other 2n3055s but they are of different brands. I want to finish the project cheap so I got the Toshiba from the "@bay" (less than $1 a piece) because I want all four of them same brand. They tested out with low hFE, and I plan on changing them once everything else is completed. I had to tweak the Iq, keeping it at around 1A. To my surprise, they sound very good (but I think the antique Motorola sounds a little better). So, I'm a bit slow to change them to 15003.

In fact, the amp sounds very musical and much better than my Musical Fidelity A1 with preamp mods. It is the best sounding amp I have ever built or own (haven't build any tube amps yet, but have three old tube radios and love them). In this regards, I really have no complaints with the fake Toshiba.
 
Last edited:
Man...I'm feeling the pressure here !

I take them as fake since the Toshibas have been largely discussed here before and most believe Toshibas are fake, especially those not from well known source. I bought Toshibas from the _bay before for repairing bad output transistors. They blew once I put them in. This time, I bought them from the _bay again so I take them as fake. Especially the low hFE makes me think they are not really 2n3055, but probably something else failed manufacturing QC. Probably someone bought them in bulk and stamp them with Toshiba 2n3055. Since everyone is saying the transistors in JLH are not that critical so I believe the good sound is coming from the design rather than the component. Enough spelling tonight, got to get a "drink".....
 
Just to report that I finally got a chance to swap out all the Toshiba 2N3055 to MJ15003 today. Life was very busy....

The result is very obvious. The amp becomes much more "liquid" and more pleasant to my ears. Female vocals apparently sweetened. Bass sounds more defined and stronger. It was a direct swap and just a few turns on the trimmer for output DC offset. The Iq is at around 1A now.

Next upgrade is to change out the phase split transistor to 2sc3421 then the input to 2sa970 as suggested by Tim Andrew.

Will report result once I can get to it........

- AC
 
Guess you woke me up !

Sorry, no update. Transistors still not replaced. LOL. Have built a couple of Class-D and tube preamps since. The JLH has been in storage reason being it uses lot of juice and heats up the room. Probably will get it out next winter (here in FL, no winter actually).
 
TR3 phase split transistor finally replaced !

After 6 and a half years, I finally got to replace the phase split transistor....

Initially I used 2n4232 (TO-66 package, like a mini TO-3). I picked the highest hFE I can get from the batch which was about 50.

I found some 2n3053, which according to spec, kind of similar to 2n1711. Their hFE tested to be about 95. I swapped it in and the amp became a different amp. The tube like magic was gone and the high freq response seemed too much and I also suspect oscillation which making some slight ringing sound.

I then recall I have some 2SD600E I salvage from a broken amp. I tested them and their hFE is in 150s. I swapped them in and the tube like magic is back with other improvement. I really love the sound. Compare to using 2n4232, overall more gain and the detail is better. I especially love listening to Yo Yo Ma's "The Swan". The sound of his cello is very touching through the amplifier.

I did search the forum to see if anyone has used 2SD600E as phase split but no avail. I can conclude this is a very good sounding transistor.

I found a local surplus store selling 2n1711 and I grabbed four. But the hFE is all over the place. The highest measured 140s and the lowest around 98. If the store owner let me use my multi meter to check hFE, I may go get a few with hFE about 150 to try. Otherwise, I will stick to the 2SD600 for now.....
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.