High Hfe or Low Hfe what's the Difference

[Phend]

I have a circuit that calls for a 2N5088.
Have them with an Hfe around 510.
Also have 2N5089 with Hfe around 900.
Have a few 2N3904 all the same Hfe 282.
Some 2N3565 that go from 280 up to 400 Hfe.
And even have some old 2N5135 Hfe in the 60's.
Which one should I use, 2N5088 ?
1) What happens if I use the 2N5135 ?
2) Suppose the circuit has four transistors,
   What would the difference be using all 5089 vs 5135 ?
   A louder circuit ?
I have seen that some DIY'ers have a "favorite" ie. 2222 and just use them.
With my latest builds (sockets) I have taken some time to switch transistors.
Not much difference ?
SO what is the difference that you may hear ? (Or smell !)

[Mark Hammer]

As I understand it - and I don't understand all that much about transistors - being picky about hfe is something that allows one to simply plunk a transistor in some known location without any adjustment of surrounding established component values. If one has the know-how to tailor the surrounding component values to extract what you want out of the transistor and achieve a given result, hfe doesn't really matter all that much as long as it's within reason.

Someone please correct me if that approach is naive or msinformed.

[antonis]

I have a circuit that calls for a 2N5088.

Could you plz reveal that circuit..??

In case you speak theoretically, what should take into account is MINIMUM guaranteed h_FE..
(we don't like Base current but we have to live with it..)

P.S.
You do know that h_FE of any particular BJT isn't constant, don't you..??

[Phend]

I measure my 3904 and all 8 are 282, 5088 89 are +/- 10. So what is considered constant, thanks.
Of course a grab bag of 5088 from amzon Chinatown or Burlington could have hfe +/- 300,  that's the QC we get now.
Here is an example,



And the case of 4 transistors,

[antonis]

I measure my 3904 and all 8 are 282, 5088 89 are +/- 10. So what is considered constant, thanks.

You're welcome but I was talking for a single device under different working conditions..
(Collector current, Collector-Emiter voltage, Temperature, etc..)

Here is an example

Any BJT of h_FE equal to (or higher than) (R2//R3) / 10*(R5+R6) can be used for Q1..
//  //    //    //    //      //      //    //    //    166 (rough estimation) can be used for Q3..

Now is your turn to confirm the above..

[R.G.]

As I understand it - and I don't understand all that much about transistors - being picky about hfe is something that allows one to simply plunk a transistor in some known location without any adjustment of surrounding established component values. If one has the know-how to tailor the surrounding component values to extract what you want out of the transistor and achieve a given result, hfe doesn't really matter all that much as long as it's within reason.

Someone please correct me if that approach is naive or msinformed.

No correction needed. Now you get the old-fogey's story.
Once upon a time, transistor makers had to fight and claw for every bit of additional hfe. Transistors were so very expensive and variable that whomever designed a circuit had to tinker values to get function out of whatever transistors and hence hfe they could afford. Especially cheapo, low function circuits like the few pedals that existed would be tweaked heavily and require special part numbers. Gain and part number/type mattered a whole lot.
Fast forward half a century. Transistors are nearly free, nearly identical within a type number, and have relatively huge gains and lower variations. On top of that, electronic researchers and designers spent 50+ years making up circuits that work fine with practically any transistor as long as it has enough high frequency response and enough hfe. So the newer the circuit design and the newer the transistors, the more likely that the individual transistor hfe makes almost no difference. That is, you're completely right.

I have a circuit that calls for a 2N5088.
Have them with an Hfe around 510.
Also have 2N5089 with Hfe around 900.
Have a few 2N3904 all the same Hfe 282.
Some 2N3565 that go from 280 up to 400 Hfe.
And even have some old 2N5135 Hfe in the 60's.
Which one should I use, 2N5088 ?

Well, you could use Keen's Second Law, which states, in part "When in doubt, whip in a 2N5088."
That "law" came from the observation that nearly anywhere an NPN is needed in a pedal, the 5088 has enough gain and voltage/current to work just fine. No, a 5088 won't work everywhere, but remarkably nearly everywhere. It's low noise, linear, yada, yada, yada. I bought them in lots of 100s  to avoid stocking other parts.

In reality, most transistor-immune circuits will work OK with any device over about 100 hfe. Swap them in and play with them if you have sockets. Only in circuits that are old enough and slavishly copied enough from the dawn of transistors will have audible differences with something like a 5088 or its ilk. If the original circuit used germanium transistors, that's a whole different subject.

1) What happens if I use the 2N5135 ?

Likely lower gain and probably some  bias drift.

2) Suppose the circuit has four transistors,
   What would the difference be using all 5089 vs 5135 ?
   A louder circuit ?

Depends heavily, heavily, heavily on what the circuit details are. Maybe nothing, maybe huge if the bias is all off.

I have seen that some DIY'ers have a "favorite" ie. 2222 and just use them.
With my latest builds (sockets) I have taken some time to switch transistors.
Not much difference ?
SO what is the difference that you may hear ? (Or smell !)

Yep, humans develop favorites. See "Keen's Second Law". I wine tasting, there are cork sniffers. In audio electronics, there are resistor-, capacitor- and transistor-sniffers. And like in wine sniffing, the older the wine/circuit/transistor, the more likely it will smell different. Better? Worse? It's in the ear of the sniffer.
But for modern circuits, the exact hfe doesn't matter a whole lot for nearly all cases.

[Phend]

>180000/13800 = 8.55
>166
OK? 

[antonis]

180000/13800 = 8.55

I'd like you to introduce your elementary school teacher to me..

[Phend]

Thanks again

[PRR]

all 8 are 282, 5088 89 are +/- 10. So what is considered constant

At what conditions?

hFE always goes below 1 at some low and some high current. It effectively becomes "infinite" (leaky AF) at some high voltage, and "dead" below about 40mV. hFE tends to rise with temperature... or is that increased leakage?

I've told this before, but my prof's quizzes said "Assume beta>50". (In those days, that was the edge where you started to pay more.)

[mac]

1) What happens if I use the 2N5135 ?

EA LFO needs a gain of >29. With a speed pot going from 1k to 101k I guess low hfe transistors have less range 

2) Suppose the circuit has four transistors,
   What would the difference be using all 5089 vs 5135 ?
   A louder circuit ?

The fOXX is an old circuit, guessing again, 5135s are closer to design specs than 5089s 

mac

[BJF]

HI There,

With the words of Hepstars technician as word of advice to the guy who built the first fuzz in Sweden
-Best to use not so high gain to make the circuit work- ( yes that had to do with the actual circuit, that was made with Germanium transistors)

Generally in transistor circuits max gain is not used and there are stabilization networks to minimize variations in hfe. The excess hfe is in a way traded for stability and higher input impedance wider bandwidth, lower distortion et.c.


The best find out what the difference possibly is build a circuit with a socket and then the specific transistors can be compared.

Have fun
Bjorn Juhl

[Phend]

^ "The best find out what the difference possibly is build a circuit with a socket and then the specific transistors can be compared."

Yes, just built a Color Sound SupaSustain. An old effect from way back.
Circuit calls for BC169B transistors, which I have (or did until the wind blew them out of my hand).
Measured Hfe of 280. All four of them. They went non too impressive.
Put in the good ole 2N3904, also with Hfe of 280. (Crossed Legs with insulation)
What a difference, the circuit works as it should. Nice compressor.

[BJF]

Hi There,

Some further thoughts:


What RG says above gives a great picture and typically you can stock one transistor type in NPN and PNP in three ranges so you have Hfe 50- 100 at fairly high current and voltage for most Control and Regulation applications; and a medium transistor Hfe at 200-400 for intermediate applications and one low noise, high Hfe above 400 to make pick up amplifiers and sensors etc and just chose one that has a package and parts number that you like

As a consequence transistor makers would regularly give out a new series and new package and part number and send out data sheet to designers and manufacturers.
Especially during late 60's through 70's and somewhat onwards manufacturers would order their house numbers on parts and so while most power amps used 2N3055 and really in any application for a product of any kind needing an NPN power transistor and as an example most manufacturers had their own house number and would in service manuals list manufacturer's partnumbers but this also applied to all parts used.
if you ordered sufficient amount you could get presorted semiconductors to meet your needs.
Phillips also had custom parts made for their products. Much of this was to keep designs in doors Then parts manufacturers made same part with different partnumber.
Then of course now and then new partnumbers for "upgraded" part like in below example of NPN transistors
Then Japanese leading the edge on HiFi
brought forth extreme low noise and hugg voltage types mainly through using exceedingly clean materials and those can be hard to replace.
Special transistors were also made to push limits on possible currents and voltage
because early transistors could only work at low voltage and so there is a Hagstrom Backing Elvis model 100W 4x12 combo with a very tiny handle on top and because it uses
germanium power transistors it has a built in regulator to limit voltage seen by the transistors and all speakers are in parallel making a 2 Ohm load.
So yes definitely history of transistors is an endless story of overcoming limitations that is until transistors become obsolete.

Example from the European nomenclature and here NPN only for simplicity
BC107,108,109
BC147,148,149
BC167,168,169
BC207,208,209
BC237,238,239
BC307,308,309
BC317,318,319
BC547,548,549
just to take a few are all very similar transistors where the 7 series would be offered in Hfe class A and B; the 8 series in Hfe class A B C and 9 series in Hfe class B and C and there would be a PNP complimentary to each.

US 2N series and Japanese 2SC ( 2SA for PNP) has corresponding devices which I don't know of the top of my head but that would show up in
ECA handbooks

Now in the data sheets hfe would be specified in a graph depending on voltage and current and noise usually only for the 9 series above would be specified depending on source impedance and IC current so for instance the 9 series would have lowest noise at IC between 1 and 2 mA.
This doesn't mean that you cannot get very low noise at say 400uA only that if absolute lowest noise design center can be put here.
This type of graph is typically only presented for Low Noise types such as C suffix in European nomenclature
Hie would also be specified with 7 series having the lowest and 9 series the highest. Also generally leakage from collector to base can be ignored in modern silicon transistors and generally for music related electronics  typically hfe and hie are the most important.

Back in the day of the European nomenclature  OC transistors leakage current from collector to base is significant and must be observed in dimensioning of circuits and this applies to all germanium transistors with exception of the last generation with golden legs- when germanium transistor finally became comparable to silicon they also became eventually too expensive but they were widely used up until 1980s.
Unfortunately some people slaughter these last generation transistors in order to extract the golden legs.

For older transistors often Icbo is so high
that setting working point can be done with a single resistor to ground from base and Icbo
forms a resistance between collector and base if you like and it resembles the most stable of the typical stabilization networks

With tubes and Jfets you can set the overtones produced with Anode/ Drain load
but it is not as easy with bipolar transistors
because unlike tubes and jfets that are voltage in current out the input of bipolar is a current to give current out.
But also bandwidth specifically with newer bipolars is significantly higher.
This can be addressed with external components such as a Miller capacitor with so and so results
If your goal is to have a specific nonlinearity
from a transistor it can be easier to use a less "perfect " transistor

But there are ways to control sound partly with current and partly with B+
There are usually with any semiconductor non linearities at the ends of possible operations especially at the beginning .

Back in the days of crystal recievers
amplification in a tube could be as low as 2-4 times and then in order to get higher amplification than 4- 16 times that two in a row would give one could introduce positive feedback and trade bandwidth for more gain at the expense of raised distortion
( the inverse of negative feedback -follows from the equation of oscillators )
This technique used in early radio techniques was abandoned largely when amplifier elements got higher gain to begin with.

There are a number of composite two transistor circuits addressing shortcomings
such as low input impedance, low gain,
low bandwidth and the latter has several solutions that widens bandwidth but for audio restricting bandwidth is often necessary in order to part get desired sound at audio frequencies and partly to reduce
reception and emission of frequencies beyond human hearing.
While some of these may look interesting
the benefits for audio might not be all that
while some can add some interesting qualities such as distortion in the passband
( the for the application useful bandwidth)

There plenty two transistor circuits meant to
result in an amplifier that may have more desirable performance than a single transistor and this can be to have high input impedance and low output impedance and high gain and low noise at the same time and for this the individual transistors will not need higher hfe than 100-300.

if you make a Darlington out of two transistors with hfe 100 the theoretical
resulting gain would be 10000.

Hfe by numbers as in the higher the better is not necessarily something to strive for.

Also the most lownoise transistors ever made have a typical hfe of 300

There are other qualities more important for audio design.

So it is not all surprising that a composite amplifier made transistors not considered special in any way can have desired qualities .
On controlling sound in transistors this comes down to overload performance
transient response, distortion and bandwidth limitations and even distortion in the passband ( here meaning within the dynamic range )

More importantly because general eq is the first layer of sound that human brain distinguishes there is a huge toolset in eq in setting desired response and eq allows certain aspects to be highlighted.
For instance specific frequency ranges can be set to produce nonlinearities and so amplifier behavior can be set to incoming frequencies e g making some frequency ranges distort more than others
While a more indirect way of controlling transistor behavior it can be tremendously effective.

Oh yes I definitely get ideas from mere transistor cases and then it is question of figuring out what to do with the given set of parameters.

Have fun
Bjorn Juhl