[Soymilk]

come to think of it, i never really did any empirical test to see that my ppa was working properly after i finished building it. i just plugged stuff into it and sound came out so i assumed it works. what would be the best way to go about testing it so i know for sure? it's been sort of nagging me at the back of my mind for a while now. function generator -> ppa -> oscilloscope should do the trick, but what if i don't have access to any of those at the moment? are there any simpler tests i can do?

 

[Koen]

Quote:

OK, some additonal troubleshooting info:

1. Removed the TLE for the (dead) right channel from the circuit to verify it's working: nicely splits 14V in half. So I guess that rules out the TLE as the source of the problem.
2. Also removed the opamp for the right channel from the socket to rule this part out as the culprit. Still seeing the same odd voltages at the pads for V+ and V-:
3.     measuring 1.56V from the opamp V- pad to opamp V+ (I expect this to be around DC nput V+/V-, so around 14V in my case)
4.     0.93V from V+ pad to signal ground (no source connected)
5.     -0.6V from V- pad to signal ground
6. <SNIP>

OK, figured it out: I had already ordered 6 new rail isolation jfets (the Q3s), because was seeing strange voltages on the opamp power supply pins.
 
While reading the tweaks section of Tangent's website I found this sentence: "You may have noticed that you can jumper across all of the Q3 positions to get a single set of power rails for everything." Well, I had wondered what if..., but didn't actually know it was 100% safe to do so... So after reading this, I went ahead and jumpered the Q3 positions (first remove the Q3s!!):
 
AMP is WORKING again!!
 
So shorting the +- DC input pads across a 12V SLA blew all but one Q3.
 
Also checked why my bass boost didn't seem to work. It turns out I had a 47k resistor as R7. I have replaced those with 10k resistors (same as R4) and will do some listening tests later today!
 
For now, I am  a happy camper and rest assured once the Q3s come in the mail I will put them back in the amp!

 

[tangent]

Quote:

what would be the best way to go about testing it so i know for sure?

An easy first test is to measure the amp's quiescent current.  If it's around 100 mA (say, +/-20 mA), it's probably fine.  If it's over about 120 mA, it could be fine, but you'd have to be able to justify it with component or biasing choices.
 
Quote:

So shorting the +- DC input pads across a 12V SLA blew all but one Q3.

I still don't see how that happened, but I'm glad you figured it out.

 

[qusp]

i'm making some upgrades to my lisa III when I replace with new batteries, thought this would be a good place to ask for info since apart from some tweaks its the same amp. I want to buy a few spare output transistors from BDENT in case of mishap and possibly an entirely new set more suitable for low impedance headphones (higher current). I already have hundreds of BC550/60C. so before I pull the BD139/140 out to check for IDSS/hfe, i'm wondering if anyone knows what they might be? being that there is no bias pot in the lisa I have to get this matched pretty right I would imagine. I saw earlier in the thread that these should be matched +/-10% at 180-190ma, do you think this still holds for lisa? 
 
so the parts are BD140-16 and 139-6, this still gives a range that is too wide to be assumed matched. any clues? i'm also going to replace the EROs with some small auricap polyprops bypassed with vishay MKP and either amtrans polyprop 220pf or the unfortunately not so easy to source and not given enough daylight here; EVOX/RIFA 5mm spaced box polypropelene  film and alfoil. I really wish these were available more places and in larger values, as you just dont see fully fledged film and foil caps in such small convenient sizes and for such reasonable prices. all are metalized PP normally.
 
so any advise on other devices to try here in the output positions and with fairly low bias due to battery power? any insight into degree of matching and IDSS values, or confirmation that I should just pull a pair to find out would be appreciated.
 
thanks for reading

 

[tarking]

I have built three Pimetas (two v1s and one v2) and decided to take the next step for fun

. I am sourcing components right now. I have about all covered but I have some trouble finding Q4s (buffer bias CCS JFETs). PN4392 is not available now and wait time for more stock is veeery long. I could get PN4391 or J310 right away. The question is which one should I get. J310 is specified from 24mA to 60mA and PN4391 is specified from 50mA to 150mA. J310 is closer to PN4392 but PN4391 is cheaper .
 
Thanks for any advice!
 
EDIT: I bought three J310s from local electronics component shop (0.80€/ea. not cheap but only type available locally right away!) and measured their Idss. I got 26 mA, 31 mA and 32 mA. Are these values a) close enough each other (there is 2K Burns) and b) is this enough for 30 mA buffer bias. Sorry for asking "dumb" questions but I have not enough knowledge in electronics to understand how this works i.e. is maximum bias current same as Idss of JFET or is it amplified.

 

[tangent]

Quote:

 
Are these values a) close enough each other

Yes, and that's probably because they are from the same batch.  This is why we don't bother to talk about "close enough" in the docs: three transistors bought from the same vendor at the same time should always be close enough.
 
 
Quote:

is this enough for 30 mA buffer bias....is it amplified.

 
 
No, it's not amplified.  The Q21/Q22 and Q31/Q32 structures are called current mirrors.  It means the current through Q4 gets mirrored to go through the output transistors.
 
There's nothing magical about the 30 mA output stage bias point.  It's a common number, but in your case, I'd just go with 20 to 25 mA and be happy.  It'll still sound good.

 

[vgjako]

Quote:

Thanks for the review!
 
 
 
Your amp may be marginally stable.  I'd try two things:
 
First, measure supply current to the whole amp while you're adjusting the bias.  If current draw goes up nonlinearly with respect to current setting, it's likely oscillating, which you could verify with an oscilloscope.
 
Second, try a "normal" op-amp, one known to work well in the PPA.

Hi again,
 
I finally found out the source of my problem, it turned out that I pulled too much current from the PSU (DACT CT-102). The power supply is rated at max +/- 200 mA, but when I measured the current going through my PPA V2 using my bench PSU I only measured 150 mA when biased to 30 mA. The power loss over the darlington transistors in the PSU is at 4,4 W, the heatsinks for the darlington transistors are in other words too small, this results in an extremely high temperature. So in other words the DACT CT-102 is not capable of delivering +/- 200 mA, but maybe under ideal conditions.
 
This explains why I was seeing such a rapid increase of DC offeset when I adjusted the bias point, and maybe why the Q3's fried.
 
 

 

[tarking]

Quote:

There's nothing magical about the 30 mA output stage bias point.  It's a common number, but in your case, I'd just go with 20 to 25 mA and be happy.  It'll still sound good.

 
Ok. Thanks a lot! I bought one J310 more because 26 mA was uncomfortably close to 25 mA if I decided to need a bit more than recommended 20 mA of bias current. Now I have all three a bit above 30 mA (31 mA to 32 mA).
 
You probably saw this coming... I have another question for you... I did not get this (yes, my daughter calls me stupid nearly every day 

):
 

Quote:

I would take your set of transistors and sort them by hFE. Then take 3 neighbors at a time for the same transistor in each of the channels. They won't be as tightly matched as if you were able to take arbitrary transistors from, say, a bag of 100, but they'll be as close as you can get them.

Do you mean by this that I should take three at a time and use these three within one channel for same part type or distribute them one for each channel for same part position? I wasn't smart enough to order more than what is needed and can't get more of them without huge cost overhead (shipping costs from the US to Europe).
 
I measured hFE for the parts and got 347 - 312 for 2N5088 part and 295 - 248 for 2N5087 part.
 
This is really strange. Maybe something wrong with my meter? Edit: I used another meter and got the same result with that one too:
 
I got hFE values 130 - 133 for MJE253G (PNP) but only 67 - 90 for MJE243G (NPN). Shouldn't NPN hFE values be bigger than for PNP part?
 
Trying to learn something at the same time here! Thanks.
 

 

[Soymilk]

 
Quote:

distribute them one for each channel for same part position?

this one

 

[tarking]

I though so but wanted to verify. Thanks! Can anyone comment about the big hFE difference between MJE243 and MJE253. If I use these parts what kind of effect does it have on the performance of the amp? I really cannot match these pairs. Oh, well, should have bought more of these too I suppose.

 

[tangent]

They're different polarities, which means they must behave differently.  There's really no such thing as matched complementary pairs, never mind what the marketing copy on the first page of the datasheet says.  Look at the curves and other data on subsequent pages, and observe the wild differences.  Be sure not to be fooled by differing axes; the top two charts on page 5 of the On Semi MJE243/253 datasheet are this way, for instance.  (hFE vs Ic)
 
Match the NPNs to the NPNs, and the PNPs to the PNPs.  It's the best you can do without buying huge amounts of extras and hoping to find uncommonly strong PNPs to pair with uncommonly weak NPNs.  Even if you did that, they won't stay matched over temperature.  Again, study the datasheet...see how the curves change shape differently over temp for NPN vs PNP.

 

[tarking]

Match the NPNs to the NPNs, and the PNPs to the PNPs.  It's the best you can do without buying huge amounts of extras and hoping to find uncommonly strong PNPs to pair with uncommonly weak NPNs.  Even if you did that, they won't stay matched over temperature.  Again, study the datasheet...see how the curves change shape differently over temp for NPN vs PNP.

You missed my point. Isn't it strange that PNP part i.e. MJE253 has much higher hFE than NPN part. Almost twice as high. I checked ON semi's datasheet and that's what prompted me to ask about this. According to datasheet hFE values should be another way i.e. MJE243 hFE values should be over 100 and MJE253 hFE values under 100. I checked meter manual but it didn't help at all. I thought that maybe because pinout is different ECB vs. EBC displayed values should be interpreted differently.

 

[Soymilk]

well if the pinout's different you should be measuring across the pins you're supposed to if you arent *shrug*

 

[tarking]

Quote:

well if the pinout's different you should be measuring across the pins you're supposed to if you arent *shrug*

 
I didn't say that I measured across the wrong pins (I am pretty sure that I wasn't but I'll check again). I was just wondering how the meter is implemented i.e. could that be an explanation to this. I am planning to borrow a better meter from work to measure hFEs (Fluke).

 

[tangent]

Some datasheets aren't clear about the pinout.  It's easy to get the pinout backwards when they only give you a 2D drawing.  The On Semi one gives a nice 3D drawing, with the metal tab showing clearly, so it's much easier to get it right.
 
As for borrowing the Fluke, beware that it won't have a dedicated hFE feature.  You'll have to build up a measurement circuit of some sort.