New Build... the GEM amp

[chipwelder]

Ouch... is exactly what the 4k7 is for... there be some funny nested feedback, on schematic, but on pcb, is left out... It only happened once, but this is disheartening, will have it in the test mule racking up some hours...

OK, shucks... you think the extensive (470uF and 470nF) bypassing will make a diffs? Any ways putting orders for the lil things from RS, so if anybody has parts they want to sell, I'll send you my to-buy list prior to ordering...

[ludo]

Hmmm. I was thinking the 4700s would go in the PSU. It seems Avnet has loads of YXF after a long drought. Will try to go by there and stock up a bit today. Happy to swap if you like, or will buy the 4700s from you if you don't need them.

[chipwelder]

we could arrange something, problem is finding someting better the ESR on the yxf's are more than twice that of the Hitano's. And there is a chipamp or two I can use them in as rails.

[chipwelder]

Worked on Friday night, worked Saturday afternoon and Sunday evening... all to have the day off to go and get some parts 
The joys of working for meself...

[chipwelder]

OK so I wanna match B556 and B547's MJE's too at some stage, I guess.

So I read the tutorials on the web... and tested one or two, I have the following issues, perhaps related to the Multimeter...
1. If I measure Hfe assuming it is gain, in the little slots, the B547's test 400 odd, the B556 test 170 odd, but both count down to lower values over time... the countdown goes slower as it goes along. SO, is it not odd that the gain for complementary devices measure that differently - or is it just the residual glue on the legs of the B556's? and why does it change over time... That little battery sign bottom left, could that be an issue?
2. To measure Vbe Switch to little diode sign, and then one measures them between B-E that gives similar readings for both types...

SO am i on the right track, do I need a better multimeter, or just a new battery - I do think the latter is needed, it now fades a bit after it measured...

Cheers!

[ludo]

Here, have a dose of negativity...

When you install the transistor in the clip, it heats up to your body temp. As it cools down, gain drops and Vbe rises to what it would be at ambient temp. It takes forever to drop that final deg C because the temp difference from ambient gets smaller and smaller.  BC546/556 come in different grades that relate to current gain. A is OK, B is good, C is very good and impossible to find. Unmarked parts... well you see what you get. They could be C, or some could be. Philips parts are different from Siemens are different from Fairchild are different from Chung Wha claiming to be Philips etc. Fairchild apparently subcontracts many out to Chung Wha and some others even worse than them, or used to. [Chung Wha is a fictitious name, as are the transistors] Trouble trouble trouble. Everything is also batch dependent. Complimentary parts are usually not. BC556 and BC547 are not complimentary. I think the lower voltage ones (547) usually have more gain. The current that the meter measures the device at should be the current that you use the device at. Gain changes with collector current. The PNP and NPN gain do not of necessity change by the same degree with collector current.

One must build a jig for this that takes a lot of devices at the same time, all clamped together so they can stabilise i t o temp. That way one can get in the ball-park at least. My approach: Stuffit! Pop a pair in there and it works fine. Degenerate the mirror slightly and trim offset there. I'm certain there's a downside to that too. JFets and Mosfets are different and can do with some screening, for some things.

You need a fresh battery, absolutely

[Ampdog]

As Ludo says ......

A further problem is that 'matching' has to be done at more or less the design collector current. The shown multimeter shows gain at a constant base input current, thus if gains differ, the collector currents also differ. One could argue that for similar gains the collector currents would be similar .... but rather an unsatisfactory situation for selecting transistors, at unknown collector currents. (One can build a transistor gain meter with adjustable collector current for such purposes.)

[chipwelder]

Ag moeder!  Stuffit I'm going to try 2sa970 or something only a few make. You know I sometimes wonder why I don't just spend the money on buying new amps!!!

[GearSlave]

Chip, I've got 100 each of 546/556's here off the same batch and reel. They should be more close than anything you can get over the counter. You welcome to a couple if you want. Should be good enough to tie you over

[chipwelder]

Why does the design spec 556 and 547's... oh the 556's make a longtailed pair and the 547's drive the MJLs... so you don't match the 556 to the 547s... OK things looking up... I's a idiot...

Thanks G, I'd love some! I'll get another battery and eyeball check if it is somewhat do-able - am not going to get 10 matching pairs out of 20 in any case.... if not I'll make visit to your side... or if you visiting C-boards, you are wecome to some coffee, op hierdie stadium sommer met whiskey by...

[chipwelder]

Okie dokie, got six matched pairs B547  ... ok matched gain within one or two with the multi (Am convinced in a circuit would be best, but i'll just hold thumbs this is enough), as with the Vbe... am doing B556 at the moment... they are a pain, they be strip mounted... and unlike the B547's where if they matched gain, Vbe was close, these aren't, which I think is strange...

Ludo I am not sure I am going to dismount all your B556's so I suspect you'll have a strip of b556 with the vbe's written on them... Or else I can just take them off, replace and sellotape... but somehow I don't think sellotape and solder mix too well...

[ludo]

No worries. I'll not be doing any automated pick-n-place any time soon I usually just cut the terminals off above the sticky strip like every other small-timer

Those are CBC556B BTW, ie the very same device of unknown origin (that nobody will admit to having manufactured) that I dissed above. They do work well though.

I looked up the spec for that DT830B multimeter. It measures Hfe at 10uA base current. So at a current gain of 200-400 you'd be kind of close to the current the device will be running at in your circuit. It measures Vbe at 800uA though. So that's not going to be so close. If I'd known anything about semiconductors, I'd have been able to draw some useful conclusion from this.

Maybe you could measure Vbe with a 9V battery and a 1Meg resistor. So Battery+ on the emitter, and Battery- via 1Meg to the base for a PNP (opposite polarity for NPN.) Then measure millivolts from base to emitter pins. Croc clips to the transistor will help to keep temperature in check, sort of.

If I'm lying like an mp3 here, much obliged if someone would point it out.

[chipwelder]

Aha... more words of experience!!! well too late now, spent the not so fast times matching transistors... so yes, you have about 40 dubiously measured b556b's that i'll drop off sometime... the other 10 are not measured as i just got sick and tired... but out of 70 b556b's I got 10 0.5% dubiously matched pairs... I think I'll just let it go for now, if I run into trouble, I will match again...

Did some soldering last night, looks about like my handwriting

[chipwelder]

Really have to sort out the ceramic caps, looks awful

[GearSlave]

Did you use Positiv20 to get the boards done? They came out pretty good I must say!

Really have to sort out the ceramic caps, looks awful

I use WIMA caps exclusively with anything I build: MKS for decoupling and FKP/MKP anywhere close to signal lines. The FKP's are brilliant for small values (where you'd find ceramics) but quite bulky in larger values. They are generally the preferred choice if you can find them. When going to Kopp I try and get as many of whatever I use as I can, simply as their supplies aren't very reliable.