How long do you wait for readings to settle when measuring germaniums?

[LucifersTrip]

I've measured many for leakage over the past month using R.G.'s method.
http://www.geofex.com/Article_Folders/ffselect.htm

I use gloves and they've been stored at ~72 deg F. The numbers [before button push] usually settle after 2 - 3 minutes, but with this particular type, it's taken longer.

After getting initial readings of ~ 1.5v - 2v, I'll walk away and 10+ minutes later I'll come back to see .5v - .72v, which would be within the acceptable limits of what R.G. considers "pretty common" - "suspicious" [~200uA -300uA]

["100uA is common, 200 happens pretty often. More than 300uA means the device is suspicious, and more than 500uA I would say is bad"]

Is that too long to wait?  

Btw, regardless of this outcome, these 2N223's have been amazing as Q1 in an Orpheum.
The hfe have been in the 150-300 range.

thanx

[brett]

Hi
as you know, transistors warm up when they are in use.  In fact, they go through quite large changes, depending on how long they've been on, the Ic, the ambient temperature, the wind speed and plenty of other factors.

Waiting a while seems a good idea.  Why would 10 minutes be too long? 

Remember that it still only gives results for one set of conditions.  It'll be totally different in different circuits, for example.  And you've noticed temperature had a big effect, so don't expect the same leakage outdoors on a cold night in Anchorage and a summer night under lights in Jamaica.

I always say: If you're worried about leakage, use a Japanese and Russian device, coz most only leak 10uA.  Coz they barely leak at all I just use a regular DMM tester for my Russian stuff (GT308, GT311, GT313).  Too easy. 
cheers

[joegagan]

install monday. test wednesday.


the following week.

[darron]

i certainly wait until the next day to double check biasing..

[LucifersTrip]

Hi
as you know, transistors warm up when they are in use.  In fact, they go through quite large changes, depending on how long they've been on, the Ic, the ambient temperature, the wind speed and plenty of other factors.

Waiting a while seems a good idea.  Why would 10 minutes be too long? 

It was the anomaly, so I questioned it. It didn't act like the other 50+ germs I tested. I dread testing a bunch of these
and waiting 10+ min a pop...

Remember that it still only gives results for one set of conditions.  It'll be totally different in different circuits, for example.  And you've noticed temperature had a big effect, so don't expect the same leakage outdoors on a cold night in Anchorage and a summer night under lights in Jamaica.

Yes, I'm well aware. I'm just using em for simple vintage fuzzes for home recording. I tried them in a Fuzz Face and
once the effect was engaged, it took a longer time for Q2's collector voltage to steady than with the other transistors
I used...There's probably a correlation?

I always say: If you're worried about leakage, use a Japanese and Russian device, coz most only leak 10uA.  Coz they barely leak at all I just use a regular DMM tester for my Russian stuff (GT308, GT311, GT313).  Too easy. 
cheers

haha...Yes, no challenge in that.

thanx

[LucifersTrip]

i certainly wait until the next day to double check biasing..

And the quality shows.  You must have a station where you can check multiple transistors at the same time

thanx

[darron]

i certainly wait until the next day to double check biasing..

And the quality shows.  You must have a station where you can check multiple transistors at the same time

thanx

well this is my friend:

http://www.altronics.com.au/index.asp?area=item&id=Q2100

following that i have a socketed booster for auditing noise. then everything gets retested the next day for a bit of shifting probably from my iron temp...

[LucifersTrip]

well this is my friend:

http://www.altronics.com.au/index.asp?area=item&id=Q2100

Yes, I've read great things about it on this forum, but haven't really looked into it yet.  I just read about it's capabilities in the link you sent. Just a few questions....When you hit "test", does it tell you if a transistor or diode is ge or si [or do you just make an educated guess from the higher leakage {for tran} results]?  Is the ge leakage accurate? Did you compare the leakage results with R.G.'s test?

I'll highly consider buying one.

thanx again

[darron]

You just hit test and it tells you what you plugged in, from an LED to a JFET to a Ge, to Si, Darlington pair etc.


I haven't made R.G.'s tester so I can't compare. I think my friend said the numbers didn't quite match up? The Leakage it reads out is very friendly though to help you estimate how it might bias and what sort of noise to expect. There's still an unsettled debate I think as to weather the 'Gain' it reads out deducts the leakage. The only thing that makes me suspicious is that I just about always notice higher gain transistors also have proportionately higher leakage on AC128s (new and old sort). I've never regretted it...

[jrod]

You can get the tester straight from Peak and it's cheaper: http://www.peakelec.co.uk/acatalog/jz_dca55.html

They recently dropped the price, too! 

[darron]

Oh. I didn't pay retail but will certainly spread the word! Thanks.

[amptramp]

I hate to be the skunk at the garden party here, but if the characteristics of a device keep changing after several minutes, it is simply too unstable to use in a pedal that may be used at any temperature.  If you are measuring hfe, it will change (probably increase) due to heating.  It should move asymptotically toward some value and not wander all over the place.  If the leakage current keeps changing, cut the leads off the device just to keep you from being tempted to use it.

Think of driving all day to Las Vegas with no air conditioning or, as I saw happen with one group, they came in late out of the cold and had to set up immediately because they had already changed the order of the four groups that were playing that night.  There was no opportunity to warm up the device.

[darron]

Some circuits aren't so fussy. HFE could easilly go up/down by 30 just from handling it. I have a draw labelled 'leaky bitches' so that I could maybe make buffers out of them one day (yeah right).

Always use a trimmer somehow for bias, then you have a less limited options for transistor selection. Maybe make a heater control element for your device :p

You take the good with the bad. I love my germanium collection of old mullards and raytheons etc. It becomes apparent very fast why the world went silicon!

[mac]

I noticed that the more leakage the more I have to wait.

Readings of 2n1114, 2sd72k taken on summer and winter,
163, 300ua  -  150, 180ua
100, 590ua  -  76, 250ua

mac

[PRR]

> How long do you wait for readings to settle...?

How long do you wait to play?

A few minutes, maybe. Next-day, no.

Leakage will INcrease due to self-heating, but R.G.'s plan (which is typical for small Ge amps) puts only 8mW in the transistor, heat should be small, maybe 2:1 INcrease of leakage.

You are seeing a DEcrease? I suspect your transistors have gone leaky in another sense: air and moisture have got in past the seals and contaminated the surface. Package leakage was a never-ending problem until recent decades.

I can't find 2N223 in my old books, so it must be VERY old.

Try this. Build a burn-in rack with several sockets, 9V, 2.2K and 2Meg each. Let some parts cook for a week. (A 9v battery may not last a week; a 9V-15V supply is safe for burn-in.) Test and see if they act more stable. Test again in a day and a week, see if they stay stable or can't be trusted.

> have been amazing as Q1 in an Orpheum

That is, of course, the only true test.

However Orpheum with 9V and 22K can't take more than 400uA Iceo before Q1 saturates, gain cuts-out. That's 1V in R.G.s fist test stage. If you get <1V before you could finish tuning and introducing the band, it should be good enough for the stage.

[LucifersTrip]

You are seeing a DEcrease?

Yes, after I insert most of the transistors there's a small jump for ~ 15 - 30 seconds, then a slow drop for a couple
minutes until they stabilize. The 223's took a lot longer

I can't find 2N223 in my old books, so it must be VERY old.

Yes, they're Philco, I believe surface barrier, most likely 50's to early 60's...the ultimate in mojo...solid gold.  Solid gold case with gold legs:
http://transistor-spravochnik.ru/description/2n223/1839

Try this. Build a burn-in rack with several sockets, 9V, 2.2K and 2Meg each. Let some parts cook for a week. (A 9v battery may not last a week; a 9V-15V supply is safe for burn-in.) Test and see if they act more stable. Test again in a day and a week, see if they stay stable or can't be trusted.

Thanx much...I will consider that when I have more time

[mac]

You are seeing a DEcrease? I suspect your transistors have gone leaky in another sense: air and moisture have got in past the seals and contaminated the surface. Package leakage was a never-ending problem until recent decades.

Freaky.
I have lots of germs still in good shape. How to safely storage them to avoid this problem?
I put the head of the transistors in my pedals into a bath of epoxy painting. Transistors are more isolated from temperature, air and moisture.
I'm thinking of re-sealing all my germs with epoxy or some kind of barnish.

mac

[PRR]

> I believe surface barrier

Why do you say that? The package? That was a fairly common case. 

Man, you sent me back to the 1953 IRE papers.

The russian datasheet says 200KHz and 250mW.

Surface Barrier is very small active area, good for 50MHz (50,000KHz), and probably not many milliWatts. Fat alloy parts were often 50mW, so it seems likely that Surface Barrier parts would be listed as lower mW.

[tubelectron]

Hi All,

Measuring Ge Q's is somewhat problematic (I prefer measure Si or tubes), and usually, most of the DMM are unable to do that correctly because of the Ice0 leakage of the Ge devices, which falses the measurement process (based on NO Ice0 leak, like on Si). Even my transistor meter DUOYI DY294 EN V3, which is a good device by the way, is not designed for correct Ge testing.

If we speak here about low power / small signal Ge transistors like AC128, 2N527, etc... :

I rely on R.G. Keen Ice0 leak's values, as I found them in correlation with the sound results in fuzz pedal tests I have made. Considering a real Hfe at 80-100 at Ib=10µA : 100µA or less is good, 100-200 is correct, 200-300 is acceptable if you have not better, 300-500 is suspicious, over 500 is usually to avoid. Of course, the lower the real Hfe, the higher the Ice0 leak = the worse is the Q. It means if the Hfe is low, then the Ice0 must be low unless it is faulty.

I made myself my test instrument, using external DMM's and regulated PSU, based on the classical measurement circuits you find in data books or electronics/student books. Here it tests the Ice0 of an AC128, supplied by VCE=9VDC, showing a leak at 139,5µA (the reading of the other meter is discarded in the Ice0 test) . Completing the test showed a real Hfe at 120 (>100) , so this AC120 was considered "good" :



And, of course, I need to wait for stabilization when testing a Ge device. That means that if the Q remains unstable, it is discarded.
When you put the Q in the test instrument, it draw heat from your fingers and so on. If you blow on it, it even will draw heat too, if you take it in a pair of flat pliers, it will loss heat, etc...

Usually, I found this for a "premium specimen" (R.G. Keen) of 100µA leak :
- When I put it, I find around Ice0 = 200-250µA,
- 2-3 minutes later, it reads 120-150µA,
- 10 minutes later it reads 80-100µA,
- 1 hour later, if I wait for, it will read around 80.
- A complete day of waiting doesn't show much more plus/minus variations.

So I consider 10 minutes as a satisfactory waiting time for stabilization of the DUT, but after 2-3 minutes you must already see a significant decrease in Ice0 leak, as stated.

A+!

[LucifersTrip]

> I believe surface barrier

Why do you say that? The package? That was a fairly common case.

On quick inspection, the 223 was very close to the surface barriers pictured at The Transistor Museum. It's similar to the
solid gold one on the left, complete with red stripe:


On second look, the package is the same as from the Russian site I linked to:


It's has a flat top and is not as thin and bullet-like like the sb ones pictured...so I don't know what type it is.
I have not found any info on it (except for the Russian site) & none listed for sale anywhere. I've not been around here long, but have not seen any solid gold ge transistors pictured in builds yet...I'm sure there are.

Man, you sent me back to the 1953 IRE papers.

It is truly a great site for old transistors, eclipsing datasheetcatalog.com & alldatasheet.com. I looked up 10-15  old transistors that were not listed at those two and ALL were listed at the Russian site. In addition, it actually has a clear (and usually accurate pic of the package).