Choosing substitute transistors, diodes

[MuseChaser]

I'm currently working on an under-performing left channel of a Dynaco SCA80Q I built as a high school kid back in the late 70s. I can follow directions, and have built quite a few kits (Dynaco, Hafler, Audio Concepts speakers, and most recently an Akitika PR102 preamp), and have done a lot of board repopulating and replaced all of the large caps on the Dynaco in the past month or so.  However, I don't know much about electronic circuit analysis or theory, although I'm making a concerted effort to learn, and have purchased a Siglent SDS1202X-E and am slowly getting comfortable with it as my first introduction to scope usage.  For anyone interested or willing to comment or help with the Dynaco specifically, this thread details the issue I'm trying to pinpoint...

https://audiokarma.org/forums/index.php?threads/dynaco-sca80q-issue-s.988842/

In general, I'd like to know more about choosing appropriate substitute transistors.  I've downloaded datasheets transistors from alltransistors.com and used their substitution recommendation pages, and have also read various websites for help with this, but many of them say things like, "choose close to the value of your current transistor," without specifying exactly what "close" means, or saying that it's OK to exceed a certain value, but not by how much.  I've gleaned that exceeding the Ft of an existing transistor can lead to oscillation in a circuit not designed for a transistor with a higher limiting value, but confess I still don't really understand what is OK in terms of a lot of the other specifications and what isn't, nor the hierarchy of importance of each specification match.

The transistor I'm having trouble matching is Q1 on PC-18 of they Dynaco, specified as a BC108A transistor, "BC108A 170-2@ Beta @ 2 ma" in the manual.  I've attached a datasheet if it's of any use.  The BC108A is not available on digikey, nor are any of the close to identical matches shown on the alltransistor substitute recommendations.  What matters? What doesn't?  How close does "close" have to be? What would be a good, available substitute for a BC108A used in the Dynaco circuit?

Another source of embarrassment - diodes D2 and D3 on the same board are spec'd as silicon diodes, "0.8 volt max. drop @ 140 ma." I can't find a single "silicon diode" on digikey... what the heck am I doing wrong? Same thing as a "standard" diode? Looking at some "standard" diodes, I can't tell what spec I should be looking at to match the Dynaco manual description.  Can anyone offer guidance there, too, please?

Thanks for any help... much appreciated.

[jmelson]

Mouser has the BC108C, probably a higher-rated version of the BC108A.

https://www.mouser.com/ProductDetail/Central-Semiconductor/BC108C-PBFREE?qs=OlC7AqGiEDncoMizMgeODg%3D%3D&mgh=1

Practically all diodes today are "silicon".  But, there used to be germainium diodes, and now you can get Schottky diodes with lower forward voltage drop.  Also, Zener diodes have controlled breakdown in the reverse direction, and are often used for voltage regulation.  So, avoid Schottky and Zener diodes.  The diode data sheet should provide forward voltage drop, but not necessarily at the exact current you want to match.
Jon

[MuseChaser]

Thanks for the reply.  I had seen the BC108C, but didn't know if exceeding the Hfe of the BC108A, list as 120 by alltransistors , up to 420 for the BC180C was OK or not.  Is "Hfe" the same thing as the beta range given in the Dynaco manual, "BC108A 170-260 Beta @ 2 ma?"

Please forgive my ignorance, and I appreciate your help.

[bob91343]

You are going about this backwards.  Finding a substitute and wondering if the differences are important isn't the way to go.

The transistor serves a circuit function and must behave in a predictable manner to accomplish that.  In order to decide what will work, one needs to understand the circuit.  Most circuits are straightforward and can be analyzed.  When you do so, you can decide what parameters are important and thus will be able to search for a part that meets those parameters.

There is no substitute for knowing how stuff works.  The original designer had a reason to select a particular part but you have no way to find out his reasons.  So you must create your own reasons.  His may have been cost, or excess inventory, unrelated to performance.

As for diodes, I have so many 1N4148 diodes that I think of that type first when replacing one.  It's pretty generic and cheap and will work in many circuits.  If I remember correctly it's rated for 100 V and 1 A and so should work for many applicatons.

[Kleinstein]

The BC108A is relatively easy. This is / was a standard small NPN in TO18 case. It is electrically nearly identical to a BC548A or BC238A as the naming iteration before, but in a TO92 case. One may need a little bending of the leads, but not to bad.  So unless you really need / want the metal case this is an easy replacement.

With the BC... series the letter at the end is usually the gain range, though a few types also use an extra number, like -16 / -25 / -40.

[MuseChaser]

You are going about this backwards.  Finding a substitute and wondering if the differences are important isn't the way to go.

The transistor serves a circuit function and must behave in a predictable manner to accomplish that.  In order to decide what will work, one needs to understand the circuit.  Most circuits are straightforward and can be analyzed.  When you do so, you can decide what parameters are important and thus will be able to search for a part that meets those parameters.

There is no substitute for knowing how stuff works.  The original designer had a reason to select a particular part but you have no way to find out his reasons.  So you must create your own reasons.  His may have been cost, or excess inventory, unrelated to performance.

As for diodes, I have so many 1N4148 diodes that I think of that type first when replacing one.  It's pretty generic and cheap and will work in many circuits.  If I remember correctly it's rated for 100 V and 1 A and so should work for many applicatons.

I understand and agree with your point, and have been dutifully, slowly, and painstakingly working my way through various EE texts, with varied success, including Horotwitz/Hill's AoE, Hayt's Circuit Analysis, and others.... but I'm hoping that I can, concurrently, get this Dynaco repaired in the meantime.  It's going to take me QUITE a long time to get to the point where I know everything you suggest (correctly) that I need/should know in order to make a decision as to what transistor would be better than what the manufacturer/original designer had intended.  Until I get to that point, I need to rely on the manufacturer's recommendations and the help and kindness of others. 

Here's the schematic, if you or anyone else would like to make a recommendation based upon analysis of the circuit...this is the amplifier board, and eyelet 12 is the input from the preamp board.

[free_electron]

bc108 is essentially a 2n2222

[MuseChaser]

The BC108A is relatively easy. This is / was a standard small NPN in TO18 case. It is electrically nearly identical to a BC548A or BC238A as the naming iteration before, but in a TO92 case. One may need a little bending of the leads, but not to bad.  So unless you really need / want the metal case this is an easy replacement.

With the BC... series the letter at the end is usually the gain range, though a few types also use an extra number, like -16 / -25 / -40.

Thanks for the information.  Another really basic, ignorant question, and thanks in advance for the patience...

Would an extended gain range transistor still behave the same in the circuit, i.e.., for a given input it would behave the same... but one with increased gain range would simply be able to handle greater signals in other more powerful circuits? Something like that? 

About the only theory I've covered so far is Ohm's and Kirchhoff's laws, voltage and current divider circuits, Thevenin equivalents, started nodal analysis, but all so far just with power sources and resistors.  SOOoo much to learn.

[Kleinstein]

It depends on the circuit, if a transistor with higher gain can be used. Often a higher gain is beneficial, but a few old circuits depend on the gain for the bias point (not good, but ot may safe a few parts). Other down sides are more early effect (dependence of the transitor gain on the collector voltage) and maybe slightly more leakage.

In the given circuit higher gain is actually a good thing. So a BC548C or its high voltage brother BC547C would be a good choice.
The 2N2222 is the same TO18 case, but a slightly larger chip and thus higher capacitoance. It should still work as well.

The circuit is a rather basic audio-amplifier, with a slight twist with the AC coupled output stage. By todays standard it would likely have quite some distortion and the choice of output transistors is a bit strange by to days standard, as they are very slow.
This is not a circuit really worth to replicate - maybe repair if needed.

The diodes D2 and D3 part is somewhat strange - and depends on what is connected to the external pin 6. Connecting pins 6 and 8 would make it a class B and the diodes would have no effect. If there is a positive side current source connected it could make the amplier a poor border-line class AB if the diodes have a rather high forward voltage. Otherwise it may still be class B with slightly reduced distortion.

[MuseChaser]

Thank you VERY much, Kleinstein, for taking the time to look at and comment on the circuit! That helps a great deal.

Here's a description of the design copied from the manual...



If I understand the description correctly (a big "if"...), Q1 and Q2 are a class A amp feeding Q3 through Q6 which operate as class B. Eyelet 6 connects to large 400 ohm 7W power resistors. D2 and D3 are open due to forced current from R16 and R17 shown here....



When the current from Q4 reaches the same current level, D2 and D3 stop conducting, which opens D1, shorting the input of Q3 amd Q4, amd outputs of Q5 amd Q6....kind of a strange protection circuit?

As I stated in the thread over on the other forum, scope traces are pretty much identical between channels at the preamp outputs, and on eyelets 12, 11, 10, and 9 on the PC-18 amp boards. At eyelet 8, they diverge greatly and the audible differences between the channels (left channel softer, and restricted sounding frequency response) becomes clearly visible on the scope as reduced amplitude in the left channel trace, and clipping of the negative portion of the wave once the volume is advanced to halfway and beyond.

What would that indicate to you? Any other tests you would suggest to unequivocally identify any bad or incorrect components? All of the caps on the board are new.

Thanks again!

[bob91343]

I'm unsure what all this commotion is about.  I suggest you slap in a 2N2222 or similar and see if it works.  If it sounds good, you are okay.  You won't blow up anythiing, no matter what.  For higher gain, a 2N2484 might be the one.

[Kleinstein]

With early clipping, the DC level at the eyelets 8 and the collector of Q2 should be checked. With no input signal the votlage at both should be at about half the supply. With some AC (e.g. sine, amplitude so that it does not clip) at the input the DC level should not shift much.

I don't think D2,D3 have anything to do with protection. It is more like an odd way to reduce the cross over gap in class B. From the transistors there are 3 BE junctions to cause a dead zone and the 2 diode subtract from that voltage. So the dead zone for the class B is 1 BE drop or maybe a little less depending on the diodes.
The more normal position for such diodes is between the base points of Q3 and Q4 and without that extra AC coupling.
It may be an odd circuit to circumvent some patent.

[TimFox]

The 2N2484 was usually specified for lower noise applications, the 2N2222 is general purpose.  Both have metal cases (TO-18).
Mouser has 2N2484 from Central Semiconductors in stock for $2.48 in singles, $2.23 at quantity 10.
I have had good luck with 2N2484s, dating back to when the original manufacturers made them.

[magic]

That's some interesting way of biasing an output stage

Datasheets indicate that some 1N4001~4007 could fit 0.8V @ 140mA, but 1N4148 won't cut the mustard.
Too much voltage dropped on those diodes may overbias the output stage with unpleasant consequences.

TO92 with EBC pinout should fit into BC108 footprint with little effort.

[TimFox]

The diode spec he quoted is 0.8 V max at 140 mA.  The datasheet typical values for 1N4148 are higher than 1 V at 140 mA.  Typical for an 1N4001 is 0.8 V at just over 100 mA.
An 1N5400 diode (3A rating) does 0.8 V typ at 900 mA.
These are all "cold" values.  An 1N4148 at 140 mA will heat up a bit, and the voltage at that current will fall.

[TimFox]

The numbers I quoted above for forward voltage are specified for junction temperature of 25^o C, but the diode will heat up with 140 mA through it, which reduces the forward voltage.
During the second intermission of the hockey game, I made a quick measurement on a normal 1N4148 diode, mounted between the spring clips (1.5 inches inside edge to inside edge) of a Grayhill component holder plugged into a DVM.
At 140.0 mA DC, the voltage (after warmup) is 0.893 V.
I'll try some higher-current diodes tomorrow.

[TimFox]

After posting the above, I thought maybe the 0.8 V max at 140 mA was for a 1N645 diode, which was more popular back then.  Its datasheet is sparse, but might be consistent with that value.
Unfortunately, I don't have any 1N645s in stock, but I checked two other diodes of that vintage:  1N459A and 1N457.
The 1N459A was slightly lower than yesterday's measurement on a 1N4148:  0.849 V at 140 mA (vs 0.893 V for the 1N4148), both after warmup.
Higher current devices:
1N4007 (1 A rating):  0.801 V at 140 mA, warmed up
UF5408 (3 A rating):  0.784 V at 140 mA, warmed up

[MuseChaser]

This is incredibly helpful.  Tim, thanks so much for the thorough answers and testing.  Kleinstein, thanks for the tips for further voltage testing points.  I'll follow through on that soon.  VERY much appreciate the all the help and guidance.