开云体育

If your timebase fails due to 2N2207 whiskers and you don't
have spares, this will save you some head scratching.

I replaced Q161 with a 2N3906, and it didn't work.
Symptoms included no trace unless the CAL knob was
all the way CCW, and then trace starting mid-screen,
with unstable triggering.

But if you replace both Q153 and Q161, it will work
fine - provided you also swap D150 and D153.

It's all about the idle loop. If you start at, for
instance, Q153's collector and go through D150, D149,
Q153, D153, Q163, and Q161, the diode and Vbe drops
must add up to greater than 0, or Q161 will not get
enough base drive.

For some transistors, it doesn't matter: Q173, Q183,
and probably Q194 can be changed without considering
the altered Vbe.

By the way, note the seemingly-meaningless Q199.
It does play a role: it keeps the -45 supply stable
by replacing drain not taken by other transistors
when the sweep is idle.

I still have a whisker in the vertical amp. I cleared
it by tapping, but I'm keeping my eye on it.

In Haste,
Dave Wise

UPDATE - SUPERSEDES PREVIOUS RECIPE

You can change Q173, Q183, and Q194 without modifying anything else.
Q173 and Q183 need to be happy at 45V so 2N3906 is out. I suggest
PN2907, BC556, or KSA992.

If you change Q161, find the wire link on the ceramic strip between Q163
emitter and R167 + Q161 base, and replace it with a Silicon diode.

If you change Q153, change D153 from Silicon to Germanium,
and install D151 B-E protection per mod 11715 in the manual.

I replaced some 2N2207's with 2N2907A's in the sweep with no problems except that I had to change R148 from 620 to 1.8K after replacing Q153 with silicon to account for the increased Vbe. I didn't replace Q194, but I would expect R133 would have to be increased for the same reason.

I replaced the vertical output 2N2207's with 2N2907A's but now the vertical bandwidth is 5 Mhz instead of the 6 MHz listed in the specs. I suspect increased collector capacitance of the silicon over the germanium is the cause. I have done some experiments with series peaking inductors, but haven't found a value that doesn't overpeak the output.

--
Bob Haas

I can’t figure out a good order for these paragraphs, please bear with me and read to the end.

There’s just nothing like the 2N2207 where sheer agility is concerned.
When you need speed I recommend KSA992 over 2N2907A. Back when I was characterizing parts to replace the 547 unblanking transistor Q373, it was the fastest part that also had a good breakdown spec, beating the 2N2907A easily. BC556 is good too but not as good as KSA992.

I tried to burn out the whiskers, using my Heathkit Signal Tracer in “noise” mode which applies 100V through 100K. Case on one pole, B-E-C on the other. Tapping the part while energized caused a virtual symphony of crashes, which didn’t die out until I had wailed on it for some time. I haven’t tested it yet, so I don’t know whether I cured it or killed it. But I’m always interested in pushing out the borders of what we can retrofit, so even if that 2N2207 is good for another 50 years now, I tried a silicon part anyway.

You didn’t siliconize Q161, did you? I had to, a long time ago. Afterwards, I experimented with R148. At first it seemed necessary to increase it, and maybe D148 too, but when I inserted a silicon diode between Q163 emitter and Q161 base (replacing the wire link on the ceramic strip), R148/D148 became a don’t care and it worked fine with original values. It seems to me that the diode addresses Q161’s increased Vbe with the least possible repercussions elsewhere. I agree that R148 should change, but I seem to have got away without it.

The other day, Q153 failed too and I replaced it. (With 2N3906 since Vce doesn’t go beyond about 25V.) I saw trouble immediately, particularly at 10ms/div where the loop failed to hold C160 discharged. The (blanked) beam just drifted off screen and stayed there. Q153 was saturating (Vce < Vbe), so instead of trying to turn it on harder via R148, I bought some emitter headroom by replacing silicon disconnect diode D153 with germanium. I know that leakage here poses a risk of inaccurate timing, but with only 1V bias it’s not going to leak much. In practice, I observed good linearity and correct timing at all speeds. I suppose I should go through that check again with the scope in a hotbox.

Q153 is a real conundrum. On the one hand, you don’t want to change the voltage at R160/C160 because that will throw off the sweep timing. On the other hand, we want the sawtooth to idle at -1V same as usual to stay in range of the HOR POS control. On the gripping hand, we want to ensure Q153 is well turned on without saturating. It seems to me that the best path out of the trilemma is going germanium on D153. But I may change my tune after playing with the heat gun.

It seems to me that with your change to R148, you also need to change D149 to germanium or D150 to silicon (either way as long as they are the same), otherwise Q145 can’t really make headway turning on Q153; instead, most of the negative shift at Q145 collector will be wasted in D150 conducting harder, leaving you vulnerable to the loop runaway I saw. But changing the diode just drives Q153 deeper into saturation so you are relying on it still having sufficient transistor action even when the BC junction is on harder than BE. I don’t like it.

If you want, I can post a marked-up schematic showing my changes.

Dave Wise

From: [email protected] [mailto:[email protected]] On Behalf Of Bob Haas via groups.io
Sent: Tuesday, September 07, 2021 9:22 AM
To: [email protected]
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

I replaced some 2N2207's with 2N2907A's in the sweep with no problems except that I had to change R148 from 620 to 1.8K after replacing Q153 with silicon to account for the increased Vbe. I didn't replace Q194, but I would expect R133 would have to be increased for the same reason.

I replaced the vertical output 2N2207's with 2N2907A's but now the vertical bandwidth is 5 Mhz instead of the 6 MHz listed in the specs. I suspect increased collector capacitance of the silicon over the germanium is the cause. I have done some experiments with series peaking inductors, but haven't found a value that doesn't overpeak the output.

--
Bob Haas

Hi Dave,

Did you ever produce a detailed post on replacing the 2N2207s in the 547 timebases?

Thanks,

Morris

I did indeed, Morris. See “Tek 547 A Sweep display during retrace”, from 2018. You participated…

My top recommendations were, and still are, KSA992 or BC556, and enlarge the speedup cap.

Dave Wise


From: [email protected] [mailto:[email protected]] On Behalf Of Morris Odell via groups.io
Sent: Tuesday, September 07, 2021 4:37 PM
To: [email protected]
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

Hi Dave,

Did you ever produce a detailed post on replacing the 2N2207s in the 547 timebases?

Thanks,

Morris

I forgot to say, the meat is at posts 142198 and 142449.

Dave Wise

From: [email protected] [mailto:[email protected]] On Behalf Of Dave Wise via groups.io
Sent: Wednesday, September 08, 2021 8:34 AM
To: [email protected]
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

I did indeed, Morris. See “Tek 547 A Sweep display during retrace”, from 2018. You participated…

My top recommendations were, and still are, KSA992 or BC556, and enlarge the speedup cap.

Dave Wise


From: [email protected] [mailto:[email protected]] On Behalf Of Morris Odell via groups.io
Sent: Tuesday, September 07, 2021 4:37 PM
To: [email protected]
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

Hi Dave,

Did you ever produce a detailed post on replacing the 2N2207s in the 547 timebases?

Thanks,

Morris

Thanks Dave,

I do remember that thread but wondered whether you had ever condensed ti into a summary.

Cheers,

Morris

No, but the post numbers I mentioned come close. Anyway, for 547, my recommendation is still BC556+220pF or KSA992+220pF.

I will experiment with my 321A this weekend per Bob Haas’ post. (Enlarge R148.)

Dave Wise

From: [email protected] [mailto:[email protected]] On Behalf Of Morris Odell via groups.io
Sent: Thursday, September 09, 2021 12:19 AM
To: [email protected]
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

Thanks Dave,

I do remember that thread but wondered whether you had ever condensed ti into a summary.

Cheers,

Morris

Bob Haas is right, increasing R148 is better than changing D153 from Si to Ge. On the slowest sweep, R160 current is only 45uA and the 0.5uA leakage of a Ge diode is significant.
But the new value of R148 should be chosen so D148 is halfway to conduction, about -0.25V on the cathode, when sweep is 0.2ms/div.

When sweep is idle, my Q153 E/B/C voltages at 0.2ms/div are: -0.33V, -0.99V, -1.39V.
That’s with changed Q161 and adding the diode as below.

UPDATE SUPERSEDES POST 186162 Sep 6

You can change Q173 and Q183 without modifying anything else.
Q173 and Q183 need to be happy at 45V so 2N3906 is out. I suggest
2N2907A, BC556, or KSA992.

You can change Q194 if you also increase R133 from 820 ohms to I don’t know, maybe 2.2K.
You want the collector to be +8.3V during sweep.

If you change Q161, find the wire link on the ceramic strip between Q163
emitter and R167 + Q161 base, and replace it with a Silicon diode.

If you change Q153, increase R148.
The goal is to get D148 halfway to conduction, about -0.25V at the cathode when TIME/DIV is 0.2ms.
In my case (having changed Q161 and added diode), the proper value was 1.3K not 1.8K.
Also install D151 B-E protection per mod 11715 in the manual.

Dave Wise

From: David Wise
Sent: Tuesday, September 07, 2021 10:06 AM
To: [email protected]
Subject: RE: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

I can’t figure out a good order for these paragraphs, please bear with me and read to the end.

There’s just nothing like the 2N2207 where sheer agility is concerned.
When you need speed I recommend KSA992 over 2N2907A. Back when I was characterizing parts to replace the 547 unblanking transistor Q373, it was the fastest part that also had a good breakdown spec, beating the 2N2907A easily. BC556 is good too but not as good as KSA992.

I tried to burn out the whiskers, using my Heathkit Signal Tracer in “noise” mode which applies 100V through 100K. Case on one pole, B-E-C on the other. Tapping the part while energized caused a virtual symphony of crashes, which didn’t die out until I had wailed on it for some time. I haven’t tested it yet, so I don’t know whether I cured it or killed it. But I’m always interested in pushing out the borders of what we can retrofit, so even if that 2N2207 is good for another 50 years now, I tried a silicon part anyway.

You didn’t siliconize Q161, did you? I had to, a long time ago. Afterwards, I experimented with R148. At first it seemed necessary to increase it, and maybe D148 too, but when I inserted a silicon diode between Q163 emitter and Q161 base (replacing the wire link on the ceramic strip), R148/D148 became a don’t care and it worked fine with original values. It seems to me that the diode addresses Q161’s increased Vbe with the least possible repercussions elsewhere. I agree that R148 should change, but I seem to have got away without it.

The other day, Q153 failed too and I replaced it. (With 2N3906 since Vce doesn’t go beyond about 25V.) I saw trouble immediately, particularly at 10ms/div where the loop failed to hold C160 discharged. The (blanked) beam just drifted off screen and stayed there. Q153 was saturating (Vce < Vbe), so instead of trying to turn it on harder via R148, I bought some emitter headroom by replacing silicon disconnect diode D153 with germanium. I know that leakage here poses a risk of inaccurate timing, but with only 1V bias it’s not going to leak much. In practice, I observed good linearity and correct timing at all speeds. I suppose I should go through that check again with the scope in a hotbox.

Q153 is a real conundrum. On the one hand, you don’t want to change the voltage at R160/C160 because that will throw off the sweep timing. On the other hand, we want the sawtooth to idle at -1V same as usual to stay in range of the HOR POS control. On the gripping hand, we want to ensure Q153 is well turned on without saturating. It seems to me that the best path out of the trilemma is going germanium on D153. But I may change my tune after playing with the heat gun.

It seems to me that with your change to R148, you also need to change D149 to germanium or D150 to silicon (either way as long as they are the same), otherwise Q145 can’t really make headway turning on Q153; instead, most of the negative shift at Q145 collector will be wasted in D150 conducting harder, leaving you vulnerable to the loop runaway I saw. But changing the diode just drives Q153 deeper into saturation so you are relying on it still having sufficient transistor action even when the BC junction is on harder than BE. I don’t like it.

If you want, I can post a marked-up schematic showing my changes.

Dave Wise

From: [email protected] [mailto:[email protected]] On Behalf Of Bob Haas via groups.io
Sent: Tuesday, September 07, 2021 9:22 AM
To: [email protected]
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

I replaced some 2N2207's with 2N2907A's in the sweep with no problems except that I had to change R148 from 620 to 1.8K after replacing Q153 with silicon to account for the increased Vbe. I didn't replace Q194, but I would expect R133 would have to be increased for the same reason.

I replaced the vertical output 2N2207's with 2N2907A's but now the vertical bandwidth is 5 Mhz instead of the 6 MHz listed in the specs. I suspect increased collector capacitance of the silicon over the germanium is the cause. I have done some experiments with series peaking inductors, but haven't found a value that doesn't overpeak the output.

--
Bob Haas

Hi all,

Forgive for reviving this older post, but It’s the closest and first I found regarding the 2N2207’s in a 547.
I found that the issue I have with my timebase is probably because both Q554 and Q564 are bad, as well as Q424. They are actually not 2N2207’s, but all 3 are marked ??AF118 4ms??. I ordered a dozen AF118 from my local shop, but they’ll only arrive sometime next week. Meanwhile, I have a few 2N2907A’s. Could those do the trick in these positions? Any resistor change to expect? I’d like to see if these could work.

Thanks in advance.

From the 547 mod summary: At S/N 14710, Q424, Q554, Q564 were changed from 151-0063-00 to 151-0322-00 (2N4890) “to improve availability”. Heat sinks removed because no longer needed.
They did not change the bias.
Based on my investigation of substitutes for Q173, Q373, and Q184, I’d recommend KSA992 or BC556 instead of the comparatively slow 2N2907. Watch the pinout, it’s different.

HTH,
Dave Wise

From: [email protected] <[email protected]> On Behalf Of Stephen via groups.io
Sent: Wednesday, April 20, 2022 9:23 AM
To: [email protected]
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

Hi all,

Forgive for reviving this older post, but It’s the closest and first I found regarding the 2N2207’s in a 547.
I found that the issue I have with my timebase is probably because both Q554 and Q564 are bad, as well as Q424. They are actually not 2N2207’s, but all 3 are marked ? AF118 4ms ?. I ordered a dozen AF118 from my local shop, but they’ll only arrive sometime next week. Meanwhile, I have a few 2N2907A’s. Could those do the trick in these positions? Any resistor change to expect? I’d like to see if these could work.

Thanks in advance.

Hi Dave,

I'm a bit confused about your comment re: slow 2N2907. That part has over twice the ft and half the output capacitance of the original transistor. A 2N3906 would do even better (still lower capacitance, somewhat higher ft). There might be other reasons to disfavor these parts, but I don't think relative slowness is one of them.

--Cheers
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070

Thanks for making me work harder, Tom.

Some spec sheets say 13pF but others say 2.3pF. I believe the latter.

Cob was more important than fT in Q184/Q373 because it is switching, with a wide output swing subject to Miller effect. fT is more important for Q554/Q564. Tek replaced those - but not Q173/Q184/Q373 - with 2N4890.
Cob is of medium importance on Q554/Q564 – the base has low-impedance drive so Miller effect is not happening.
2N4890 fT is 100MHz, BVceo is 40V and Cob is 15pF.

If you believe the schematic annotations in the 547 manual, Q554 can see 65V C-E. Since the base is not floating around, BVcbo (60V) is more important than BVceo, but 2N3906 BVcbo is only 40V and you run the risk of breakdown.

Q424 is in Delay Pickoff with a tunnel diode on its base; Miller is important and I’m surprised 2N4890 worked well. It seems to me that 2N3906 would be fine here, or KSA992.

When all is said and done, if I ever replace Q554/Q564 in my own 547, I will try KSA992 or BC556 before 2N3906, unless I selected the latter for breakdown.

For what it’s worth, 2N4890 is in stock at Boca Semiconductor, $2 each plus $5 shipping.

Dave Wise

From: [email protected] <[email protected]> On Behalf Of Tom Lee via groups.io
Sent: Wednesday, April 20, 2022 12:16 PM
To: [email protected]
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

Hi Dave,

I'm a bit confused about your comment re: slow 2N2907. That part has
over twice the ft and half the output capacitance of the original
transistor. A 2N3906 would do even better (still lower capacitance,
somewhat higher ft). There might be other reasons to disfavor these
parts, but I don't think relative slowness is one of them.

--Cheers
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
<>

On 4/20/2022 12:00, Dave Wise wrote:

From the 547 mod summary: At S/N 14710, Q424, Q554, Q564 were changed from 151-0063-00 to 151-0322-00 (2N4890) “to improve availability”. Heat sinks removed because no longer needed.
They did not change the bias.
Based on my investigation of substitutes for Q173, Q373, and Q184, I’d recommend KSA992 or BC556 instead of the comparatively slow 2N2907. Watch the pinout, it’s different.

HTH,
Dave Wise

From: [email protected]<mailto:[email protected]> <[email protected]<mailto:[email protected]>> On Behalf Of Stephen via groups.io
Sent: Wednesday, April 20, 2022 9:23 AM
To: [email protected]<mailto:[email protected]>
Subject: Re: [TekScopes] 321A 2N2207 transistor replacement: Ge-Si replacement sensitivity in the timebase

Hi all,

Forgive for reviving this older post, but It’s the closest and first I found regarding the 2N2207’s in a 547.
I found that the issue I have with my timebase is probably because both Q554 and Q564 are bad, as well as Q424. They are actually not 2N2207’s, but all 3 are marked ? AF118 4ms ?. I ordered a dozen AF118 from my local shop, but they’ll only arrive sometime next week. Meanwhile, I have a few 2N2907A’s. Could those do the trick in these positions? Any resistor change to expect? I’d like to see if these could work.

Thanks in advance.

Thanks for doing the research, Dave! I had not seen a datasheet with a 2.3pF value, but seeing that the ICmax spec is 50mA, I'm pretty sure you're right. The 3906 typically has around that value of Cob, but the breakdown voltage requirements rule out its use.

--Cheers,
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070

As it turns out, all the 6 AF118’s are bad. 2 are completely dead, and the 4 others show a diode across EC.

KSA992 are not available in my area, and I don’t feel like ordering them online. BC556B are available though. Is there anything wrong with using the AF118 I ordered? They were original to my unit after all.
Thanks Dave, but ordering 2N4890 from Boca would be quite expensive from Europe. Not a good option.

I tried replacing all 6 of them with 2n2907A. After doing so, not only the problem I have did I not disappear, but another one popped up. Sparkles spots are randomly appearing on the trace. They were not present before. Not a good sign.
To summarize, the original issue was no ALT whatsoever, and no Delayed Time base.
When in regular ALT position, only A is displayed.

I can easily go pickup some BC556B’s tomorrow.

I believe it’s worth mentioning that my unit has a 70000 serial number, and needs to follow the calibration procedures of units with serial numbers below 6739. It’s a rather early one.
For what it’s worth…