Re: Fix or Part Out a Tek 475A
I once went to a course in London, Frequency Synthesis by a Robert Dixon, an American. He said
- call me Bob, Mr. Dixon is my father (sorry folks, I could not help it).
G?ran
|
Re: (OT) Where to go for 70s IBM hardware? I'm looking for a terminal.
There are several Facebook groups that deal with buying,selling,and collecting vintage computer hardware
toggle quoted message
Show quoted text
On Nov 25, 2020, at 4:48 AM, cheater cheater <cheater00social@...> wrote:
Thanks, I'll check it out. If anyone has other suggestions, please do let
me know.
On Tue, Nov 24, 2020 at 10:19 PM Geoffrey Thomas <geoffreythomas@...>
wrote:
Hi,
Have you tried ClassicCmp?
Seems to be a real hive of info of the sort you're looking for.
Geoff.
On 24/11/2020 15:42, cheater cheater wrote:
Hi all,
sorry for the offtopic message, not sure where to go with this one, so I thought the smart heads around here might be connected somewhere. I'm
looking to get a 70s IBM terminal, either for a System/3, /6, /32, /38 (or
contemporary), 3600 FCS, a Display Station (or terminal) for the IBM 32xx,
37xx, 49xx, 52xx, 65xx, or 74xx, or an IBM 5100/5110/5120 portable
computer. I'm interested to see how they were built and would like to see
how it is to use them. If you know any good places to go other than the
website which should not be named, especially any forums that could have
enthusiasts of the same hardware, or if you know someone willing to part
with one such machine, let me know please.
Thanks!
|
Re: (OT) Where to go for 70s IBM hardware? I'm looking for a terminal.
Thanks, I'll check it out. If anyone has other suggestions, please do let me know. On Tue, Nov 24, 2020 at 10:19 PM Geoffrey Thomas <geoffreythomas@...> wrote: Hi,
Have you tried ClassicCmp?
Seems to be a real hive of info of the sort you're looking for.
Geoff.
On 24/11/2020 15:42, cheater cheater wrote:
Hi all,
sorry for the offtopic message, not sure where to go with this one, so I thought the smart heads around here might be connected somewhere. I'm
looking to get a 70s IBM terminal, either for a System/3, /6, /32, /38 (or
contemporary), 3600 FCS, a Display Station (or terminal) for the IBM 32xx,
37xx, 49xx, 52xx, 65xx, or 74xx, or an IBM 5100/5110/5120 portable
computer. I'm interested to see how they were built and would like to see
how it is to use them. If you know any good places to go other than the
website which should not be named, especially any forums that could have
enthusiasts of the same hardware, or if you know someone willing to part
with one such machine, let me know please.
Thanks!
|
Re: Fix or Part Out a Tek 475A
Hi Jeff,
Please call me Tom. :)
Some critical diode parameters are rarely discussed in EE curricula, so you have every right to be flummoxed by datasheets. It is also not infrequently the case that manufacturers want to overwhelm you with a lot of data to hide the fact that they didn't tell you the one thing you really wanted to know. One of these is "reverse recovery time". It happens that PN junction diodes don't turn off right away. Until they do, the diode maintains a 0.7V anode-to-cathode voltage, even if that means that the current reverses direction. As you might imagine that can be a Very Bad Thing(tm) in certain circumstances. The time taken for a diode to return to the non-conducting state is the reverse recovery time. This value depends on both the forward and reverse current, but under typical conditions, the 1N4148 recovers in the several-nanosecond range. The venerable 1N400x is much slower, with 5-20us being typical. The UF400x is a fast recovery version of the 1N400x series, with typical recovery times in the range of 50ns-100ns.
On top of all that, you might care about capacitance, as in any circuit that has to support a high bandwidth. Frustratingly, manufacturers often leave out that data, making it difficult to figure out whether a given diode might be an acceptable replacement. In that case, you have to guess. A good guide to guessing is to assume that the junction capacitance is 10-20pF per ampere of rated forward current. This figure applies to junction diodes, and not to Schottky or other types. (But the same rule of thumb applies to the output capacitance of a bipolar transistor, as it happens.) The logic is that both capacitance and current-handling capacity are functions of device area, so the two should be correlated. Other factors affect the relationship as well, but current handling capacity is the simplest proxy for capacitance that is quantitatively useful.
In your circuit, you're looking for a diode with a breakdown voltage of 175V and a forward current rating of 100mA. The 1A rating of a 1N4004? or UF4004 means that it can handle those two fine, but will have 10 or so times the capacitance, and so will likely be unacceptable. A Schottky unfortunately has much higher capacitance per ampere, so it is also probably not a good choice here.
After all that, I often replace these types of diodes with series combinations of 1N4148, if there's room, because I am always impatient to fix the thing now, rather than wait for the right part to get shipped to me (if I can even find a vendor of these often-obsolete components). In this circuit, two in series should suffice to replace an FDH2161. Some fussy folks would advise using a resistor in parallel with each diode to guarantee equal voltage division, but they're unnecessary here, especially if the diodes come from the same batch.
Good luck with your continued debug. You are going to get this to work!
--Cheers,
Tom
--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
toggle quoted message
Show quoted text
On 11/24/2020 23:28, Jeff Dutky wrote: Dr. Lee,
Thank you, that feels like high praise, and it means that I'm learning something. In my defense, I was going to be an EE, before I bailed to CS in my junior year, so I had a bit of exposure to some EE topics, just not enough to feel like I really understand much of anything beyond LRC circuits.
I have found another shorted diode, by the way, and am having significant trouble tracking down a data sheet for the part. The part is designated CR1367, and the service manual lists that as a Fairchild FDH2161. I can't find a data sheet for that part, but the service manual indicates merely "SEMICOND DEVICE: SILICON, 175V, 100MA" which I take to mean that the V[RRM] is 175V and the operating current is 100 mA (not sure if that's supposed to be I[F], I[FSM] or I[R]).
From the schematic it looks like CR1367 should be expected to have a reverse voltage between 95 V - 118 V, though it looks like up to 110 V could be passed through Q1358 and Q1354 to the cathode of CR1367 (but I can't figure out what conditions would cause that). Under any circumstance it does not appear to me that there should ever be a condition where the cathode voltage should exceed the anode voltage (the anode being connected directly to +110 V, and the cathode only being connected to +110 V after it passes through two transistors and another diode).
I've found a post in this same group that suggests using a "UF4004" or a Schottky diode as a substitute for a FDH2161, though it's not clear to me that the application is the same. It's also not clear to me if a UF4004 is the same as a 1N4004 (which I happen to have a supply of on hand). The diode in circuit is one of the clear orange kind, while the 1N4004 is a larger black encapsulated kind. My (possibly unfounded) impression is that the larger black encapsulated kind are meant for higher power applications.
I readily admit to being completely flummoxed by the specifications and packaging of diodes and transistors.
Any guidance would be appreciated.
-- Jeff Dutky
|
If the measured resistance is higher than that in the circuit diagram you may have a problem as it may not bypass enough current across Q1354. If you are getting replacements it would be worth putting them in, I would have tried it out and checked the temperature of Q1354 with a calibrated finger.
Simon
|
Re: Fix or Part Out a Tek 475A
On my 475 parts list it mentions TI60 or 1N647, which are general purpose diodes with 400 V reverse breakdown so a 1N4004 would replace it.
Simon
|
Simon,
my reading of the schematic says that once I have pulled Q1352, Q1354, and Q1358 from their sockets the resistors are only connected to the rest of the circuit through two diodes, one of which is reversed biased (assuming that it's not shorted, which is not a given in this case).
If the diode across the collector and base of Q1354 were shorted, however, I would expect that the resistance I read across the two resistors would be less than 15 Ohms, since some other resistance would be in parallel with them. The resistance I'm reading is higher than the marked value of the resistors (and the value indicated in the schematic), not lower.
That said, I really should check the diode CR1354 to make sure it's not also shorted, because, with my luck, it probably is.
(I was originally planning to shelve this project until the replacement transistors arrived, but then I got to thinking about why a couple of transistors might be blown, and started worrying about other components that must have failed in the same event, and realized I could test a bunch of the components if I just removed all the socketed transistors from the circuit, leaving a bunch of disconnected legs. I'm really glad I succumbed to my worry on this, because I'm finding, and learning, all kinds of things along the way)
-- Jeff Dutky
|
Re: Fix or Part Out a Tek 475A
That could explain why you have a positive voltage at TP1364. I am not an electronics engineer, but would use a 1N4004 as it looks as though it protects Q1362.
Simon
|
Re: Choosing the right replacement cap
I guess I don't quite get what the ripple current spec is actually telling me. That this is the amount of ripple current the cap can take to make the lifetime spec?
A spec with twice the ripple for a given lifetime can take more power supply noise?
A UCY2C470MPD1TD with 375mA ripple and 10000 hr lifetime can take more power supply noise than a ULD1H470MED with 190mA ripple and 10000 hr lifetime?
The UCY is cruising in the 465 HV oscillator supply while the ULD might suffer degradation despite being well within voltage and temperature specs?
How much do I care about ESR in a power supply filter?
|
Re: Fix or Part Out a Tek 475A
Dr. Lee,
Thank you, that feels like high praise, and it means that I'm learning something. In my defense, I was going to be an EE, before I bailed to CS in my junior year, so I had a bit of exposure to some EE topics, just not enough to feel like I really understand much of anything beyond LRC circuits.
I have found another shorted diode, by the way, and am having significant trouble tracking down a data sheet for the part. The part is designated CR1367, and the service manual lists that as a Fairchild FDH2161. I can't find a data sheet for that part, but the service manual indicates merely "SEMICOND DEVICE: SILICON, 175V, 100MA" which I take to mean that the V[RRM] is 175V and the operating current is 100 mA (not sure if that's supposed to be I[F], I[FSM] or I[R]).
From the schematic it looks like CR1367 should be expected to have a reverse voltage between 95 V - 118 V, though it looks like up to 110 V could be passed through Q1358 and Q1354 to the cathode of CR1367 (but I can't figure out what conditions would cause that). Under any circumstance it does not appear to me that there should ever be a condition where the cathode voltage should exceed the anode voltage (the anode being connected directly to +110 V, and the cathode only being connected to +110 V after it passes through two transistors and another diode).
I've found a post in this same group that suggests using a "UF4004" or a Schottky diode as a substitute for a FDH2161, though it's not clear to me that the application is the same. It's also not clear to me if a UF4004 is the same as a 1N4004 (which I happen to have a supply of on hand). The diode in circuit is one of the clear orange kind, while the 1N4004 is a larger black encapsulated kind. My (possibly unfounded) impression is that the larger black encapsulated kind are meant for higher power applications.
I readily admit to being completely flummoxed by the specifications and packaging of diodes and transistors.
Any guidance would be appreciated.
-- Jeff Dutky
|
Re: Choosing the right replacement cap
This was very helpful to get me started:
In trying to understand the Nichicon series Mark gave I found this:
Which linked to this:
I don't see an explicit ESR spec in the data sheets. I understand from the above that it's frequency dependent, but from Fig 1-8 it looks relatively flat. Is there a proxy for ESR?
I think ripple in this application will be a significant factor. The supply is unregulated and the purpose of the filter is to keep the 50kHz HV oscillator out of the power supply (from the Tektronix service manual). That seems like a potentially strong amount of current noise. There's a reason these keep blowing up.
How are places like Mouser and Digikey with small time onsie-twosie buyers like me? I won't be buying a real of 2000. But I think 10 is a reasonable number when the per cap price is less than $1.00. It looks like Mouser has, for example, the UBW in individual quantities. That means it's in stock and I won't be waiting on the 15 week factory lead time? If I get in the habit of buying "parts" 465s I imagine having a nice stock of these is worth it.
I won't be getting the UBW - too short a life. But now I see what the parameters are and will choose a series with high lifetime and ripple as primary drivers. I think 105C should be plenty inside a scope. Let's see, in F thats: 5/9 * C + ? Oh geez. Let me look that up.
Dave
|
Re: Choosing the right replacement cap
Mark Vincent is spot on regarding electrolytics.
Film caps - Polyester (cheaper and smaller), or Polypropylene (larger and expensive) are usually a no-brainer ultra-long-life substitute for tantalums, not considering their higher price.
One other spec which must be taken into account, is Ripple Current capability, usually specified in AC mA, in datasheets.
An unregulated or switching circuit has significant ripple, and therefore the cap's Ripple capability is more relevant, as opposed to a circuit following a voltage regulator.
Menahem Yachad
www.CondorAudio.com
|
Didn't think of that, duh.
Simon
|
I think there is a way.? If nothing else is connected at the junction of the resistors, short the ends of the combination together and measure from the shorted ends to the junction of the two resistors.? Make sure your Ohmmeter is isolated.? You should measure 3750 Ohms.? No soldering required.
Bob
toggle quoted message
Show quoted text
On Tuesday, November 24, 2020, 09:45:52 PM PST, Simon <tenareze32@...> wrote:
I think Roger¡¯s explanation of the reason for two resistors in series makes sense (better heat dissipation in a small space) as 2 x 7.5 k? = 15 k?, which is a common value. It looks as though they are there to alleviate Q1354 and reduce its C-E current. Before you change these resistors, are you sure they are out of spec.Without desoldering the junction and testing each one individually you can¡¯t be sure. You cannot accurately measure resistance in circuit. Roger has a keen eye and I would do as he suggests.
Simon
|
I think Roger¡¯s explanation of the reason for two resistors in series makes sense (better heat dissipation in a small space) as 2 x 7.5 k? = 15 k?, which is a common value. It looks as though they are there to alleviate Q1354 and reduce its C-E current. Before you change these resistors, are you sure they are out of spec.Without desoldering the junction and testing each one individually you can¡¯t be sure. You cannot accurately measure resistance in circuit. Roger has a keen eye and I would do as he suggests.
Simon
|
Re: Fix or Part Out a Tek 475A
Hi Jeff,
As you correctly surmised, a 1N4148 would indeed be perfectly fine in this case. Those two diodes are pretty much the same, with the '4152 tested to tighter tolerances. For a non-EE, you did an excellent job of analyzing the situation.
--Cheers,
Tom
--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
350 Jane Stanford Way
Stanford University
Stanford, CA 94305-4070
toggle quoted message
Show quoted text
On 11/24/2020 21:15, Jeff Dutky wrote: So I went in an desoldered CR1352 to verify that it was shorted, and it is, so now I need to replace it. It's a 1N4152, which appears to be hard to come by, but I'm seeing things on line that suggest that a a 1N4148 is a suitable replacement. Now, I'm no electrical engineer, so I don't really understand how you would select a diode for a given application (aside from knowing that a diode permits current to flow in one direction, with an approximately constant voltage drop, and that different diodes have different maximum power dissipation).
The 1N4152 is a 500 mW part, with a capacitance of 2 pF, a recovery time between 2 and 4 ns, and a forward voltage drop of 500-750 mV.
The 1N4148 is also 500 mW, with a capacitance of 4 pF, a recovery time of 4 ns, and a forward voltage drop of 1 V.
None of those specs (except for the power dissipation) sound similar to me, but again, I don't know how you would select a diode for a given application, or what the actual requirements of the diode in this application are, so maybe it's fine.
Now, I happen to have a bunch of 1N4148 diodes ready to hand, and I suppose that they're better than nothing (and better than a shorted diode), so I could just use it and hope for the best. The wattage and reverse breakdown voltage specs seem pretty good (the 1N4148 has a reverse breakdown voltage of 75-100 V, which should be fine for this circuit, and far exceeds the reverse breakdown voltage of the part that it's replacing).
(actually, just writing this I have mostly convinced myself that the 1N4148 will be fine, but only because I can't really break the scope any more than it's already broken).
If anyone is willing to try to make me less ignorant on this topic it would be appreciated.
-- Jeff Dutky
|
Milan,
Thanks for this explanation, it makes sense to me. I've noticed that component values that were common 50 years ago (or more) are not common today, so the reverse is also understandable. It was also suggested that this was an attempt to make a higher wattage part than was either available, or than would fit in the space provided. The resistors are bigger than 1/4 W ones, but the manual only specs them as 1/2 W.
Regardless, they've drifted pretty far out of spec (more than 20%) and will need to be replaced. I've ordered a selection of 1/2 W resistors that should include 7.5 K Ohm parts.
-- Jeff Dutky
|
Re: Fix or Part Out a Tek 475A
So I went in an desoldered CR1352 to verify that it was shorted, and it is, so now I need to replace it. It's a 1N4152, which appears to be hard to come by, but I'm seeing things on line that suggest that a a 1N4148 is a suitable replacement. Now, I'm no electrical engineer, so I don't really understand how you would select a diode for a given application (aside from knowing that a diode permits current to flow in one direction, with an approximately constant voltage drop, and that different diodes have different maximum power dissipation).
The 1N4152 is a 500 mW part, with a capacitance of 2 pF, a recovery time between 2 and 4 ns, and a forward voltage drop of 500-750 mV.
The 1N4148 is also 500 mW, with a capacitance of 4 pF, a recovery time of 4 ns, and a forward voltage drop of 1 V.
None of those specs (except for the power dissipation) sound similar to me, but again, I don't know how you would select a diode for a given application, or what the actual requirements of the diode in this application are, so maybe it's fine.
Now, I happen to have a bunch of 1N4148 diodes ready to hand, and I suppose that they're better than nothing (and better than a shorted diode), so I could just use it and hope for the best. The wattage and reverse breakdown voltage specs seem pretty good (the 1N4148 has a reverse breakdown voltage of 75-100 V, which should be fine for this circuit, and far exceeds the reverse breakdown voltage of the part that it's replacing).
(actually, just writing this I have mostly convinced myself that the 1N4148 will be fine, but only because I can't really break the scope any more than it's already broken).
If anyone is willing to try to make me less ignorant on this topic it would be appreciated.
-- Jeff Dutky
|
Jeff, the resistors connected end to end are an attempt to make a required resistance that was not available as a standard resistor value. Shortage of parts? Engineering change? Select in test? Who knows. I have a few of those in my 453 scope.
|
Re: Choosing the right replacement cap
Thanks Craig and Mark,
Yes, this is C1419 on a 465. Actually C1418/C1419, the two 22uF parallel caps on higher SN scopes. Mine is 29X,XXX (don't have it right in front of me at the moment). One of the two is shorted.
I started with a search on 47uF, 50v as someone in an earlier thread regarding this cap noted the marginality in this supply filter. The supply is "15v unregulated", but many note the actual value can be in the 20s. Making the 25v 22uF's right at their limit. And then adding temperature effects further stresses them/it. I felt 50v was a good "over spec" to handle the application.
So I did follow the link that came up for Mouser, and one of the selection criteria in their results is operating temperature. Considering this caps environment I selected 135+ C, just to see what it turned up. The result is a Nicicon UBW1H470MPD. It is an Aluminum Electrolytic and all of the specs look good for the application. But when I look at, for example Digikey I see other parameters. After some study I see what's meant by ESR, and understand leakage current as it applies to cap usage. I am an EE by trade, so I'm not completely ignorant. But there's a vast difference between some academic understanding and the real world application of discrete components. All my cap experience/understanding is IC parasitic cap.
One spec I don't know what to do with is "Series". I see various references to this alpha-numeric code, as Mark mentioned above. But I haven't found an explanation of it yet. A little tutorial would be helpful and appreciated. A link is fine. I have more studying to do.
I do now know better why the Nicicon is a far better choice than the cheap results on Amazon. So I feel comfortable making an order of, like, 10 of them from Mouser.
Is that way overkill? Or is it good to get a high temp cap like that? I'll try and re-read Marks answer and understand some more of what he's saying. But I hope I've clarified where I'm at better. Any further help narrowing choices and helping me understand WHY those are better choices is greatly appreciated.
Thanks,
Dave
|
G?ran Krusell
Dave Peterson