Transistor Looking Device with G21 Marking

[bostonman]

I'm working on an oscilloscope and came across a section of the circuit that has different resistance from the other two working channels.

One section of the circuit goes into a device that I think is a SOT23 package (although don't quote me) three-pin device - I assume it's a transistor. When I (blindly) measure two pins, I get an approximate 1.4v drop, but, on a good device (but still in the circuit), I get a 0.7v drop.

The marking on the component in each channel is: G21, but one or two channels seem to have a sideways 32 printed on it. I assume the 32 is a lot date code since the other devices don't have the 32.

In any case, I did a search for 'G21' and found a datasheet on alldatasheet, it indicates the G21 is just a marking, but the actual part number is: DTD113ZK

Does anyone know if I found the correct datasheet, because, according to this datasheet, one pin should be ground, and I didn't measure either of the three pins as ground.

[ataradov]

The one with sideways marking could also be DMG2301U.

The one you found is just a pre-biased transsistor, it can be used in many ways including without connecting anything to ground. You need to figure out a bit more of a schematic around those parts.

[bostonman]

The four circuits are identical (it's a four channel oscilloscope) and all the circuits have identical components in the same locations.

It's why I assumed the sideways marking was a date code, but I'll look at the surrounding circuits more. This is an extremely difficult board because it's multi-layer with blind vias. It took several hours to trace the one point (now I discovered a second point elsewhere) that was measuring incorrectly, and, when I removed all the components, I realized this device was also in the circuit.

Sadly, this scope has an obsolete IC and I'm hoping the different readings is a surrounding component rather than the IC chip.

Looking at the datasheet for a DMG2301U, it alludes that the date code would be a letter and number rather than two numbers.

[lordium]

One of my favorite sites for looking up codes:

http://chip.tomsk.ru/chip/chipdoc.nsf/vc1!readform&view=smd&cat=G&start=501&count=500

G21   2SK1958       SC-70   Small Signal FET(MOS type) Switching
G21   DMG2301U       SOT-23   P-CHANNEL ENHANCEMENT MODE MOSFET
G21   DTD113ZKA       SC59   npn digital transistor 1k0 +10k 50V 500mA
G21   MMBZ4621-V   MMBZ4621   SOT23   Small Signal Zener Diode (3.6V)
G21   DTD113ZN3       SOT-23   NPN digital bipolar transistor

so maybe last one is the likely one? Pull it off the pcb and try to measure the resistance to verify?

[bostonman]

Attached is a circuit that I've traced. I've done the best I could to trace this circuit; the board is big with blind vias. From what I can tell, this is quite accurate, however, as noted, the scope probe input section is simplified.

I'm reluctant to remove a good G21 device to avoid damaging it or it flinging off somewhere. Ideally I'd like to know, and learn, what this device is.

A slim chance this component was replaced in one or two of the channels (remember two channels out of four are bad - but only in 50ohm mode). I noticed an AD823 IC was replaced in Ch2 and Ch3 (the bad channels) - at least it looks like they were replaced because it has some solder balls along with black marking rather than shiny silver as Ch1 and Ch4 (the good channels) have.

It's possible the device(s) marked with a sideways 32 are a (wrong?) replacement. I'll try looking at the part number later to see which ones have the 32, but wanted to point that out just in case it makes a difference.

Due to them be tiny, poorly carved part numbers, reflections under the microscope, etc... it's hard to read the part numbers.

On a side note, I assume the abbreviated marking (G21 in this case) is due to limited size to print the part number, however, why does a G21 have many different part types?

[ataradov]

Pin4 of the AD736 is VSS. Given that it is a front-tend, it is likely to be a negative voltage.

So pin 3 of the device is connected to ~VCC/2 and pin 2 is connected to ~VSS/2. Both are static voltages. And pin 1 is then connected to the signal path.

The only way this makes any sense to me if this is a diode array for clamping. Try to remove a good device and measure 2-1, 1-3, 2-3. You should see a single diode drop in the first two measurements, and a double diode drop in the last.

You did not mention which pins measure 1.4 V, but it may make sense if you measured 2-3.

The reason markings are conflicting is that there is no common database, each vendor comes up with their own scheme. The goal of the markings is not to identify a completely unknown device, but to confirm that they match to the expected device.

[lordium]

just looking at the tests in your pdf I would say it's the DTD113ZN3. The first result seems like a broken one, but the second seems right, about ~1.7V and ~10k Ohm (depends on the current output of your multimeter etc)

[bostonman]

This is a huge help, but, according to the datasheet, the package dimensions don't match my measurements.

Unfortunately even if this (assumed) DTD113ZN3 is bad, the problem with this scope is an Agilent hybrid chip. I've measured several points and the area around this DTD113ZN3 measures differently. After probing more in another area, I found the AD823 IC had much lower resistance on the bad channels. My guess is that a previous owner found the same issue and attempted to replace these in hopes to fix the channels; and why they appear to have been replaced.

Sadly once I removed the IC, the circuit still measures low and all surrounding components are not contributing to the problem. The only unknown is the hybrid. Trying to find a non-counterfeit part is next to impossible, and trying to find one is harder. Also, they are not easy to replace because of a ground pad underneath.

[ataradov]

If it is a transistor, then your schematic makes no sense.

Measure the voltages on the pins of the working channels while it is powered on.

[lordium]

Yeah I take the readings and measurements with a grain of salt and read some between the lines. But Rohm makes a DTD113Z in SC-59 package (G21 marking) similar to sot-23 but a little bigger, so fits your measurement. But a transistor on some oscilloscope input seems likely (as some kind of protection I assume). And since you do have good channels to work with, it's all about how far down the rabbit hole you want to go (start swapping!).

[bostonman]

I can't disagree on this circuit not making sense, especially since I have two unknown components: the one marked G21 and a hybrid. I measured point-to-point several times, so a possibility exists I made an error in transition from measuring, to writing, to drawing the circuits, but I've checked everything several times. I made a small error on the schematic - the 100 ohm resistor is before pin 1 on the G21 - I've redrawn that section - and added the second circuit in question on the bottom. Attached is a picture of the board - I drew a red circle around G21 and two lines pointing to the vias that are connected. So I've pretty much have one pin that's correct because it's visible on the board (the 100 ohm resistor), and I am confident the two vias are connected making two pins correct.

I removed the G21 from a good channel and measured it - see attached for the updated list of measurements.

From what I can tell, the AD 823 on both Ch 2 and Ch 3 (the bad channels) was replaced by someone, and, it appears the G21 component too. My measurements led me to these two components, and, after looking at the soldering, it appears they were replaced by someone finding the same differences in measurements. I plan to replace both components in both channels, so I'm still interested in knowing what this G21 component is.

What I also noticed is the AD 823 in Ch 2 and Ch3 has a dark marking instead of a shiny silver one; and they are AD 823A instead of 823 (not sure if it's an issue, but I'll look later, just wanted to mention it). The interesting thing is the G21 in Ch 2 and Ch 3 have the sideways 32 suggesting they were replaced with the wrong component because Ch 1 and Ch4 have just a plan G21.

Attached is a picture of each G21 showing the 32 on two of them to avoid any doubt on what I'm referring to.

[bostonman]

Yeah I take the readings and measurements with a grain of salt and read some between the lines. But Rohm makes a DTD113Z in SC-59 package (G21 marking) similar to sot-23 but a little bigger, so fits your measurement. But a transistor on some oscilloscope input seems likely (as some kind of protection I assume). And since you do have good channels to work with, it's all about how far down the rabbit hole you want to go (start swapping!).

I replied to the thread before I got notification you sent a follow up - don't want to seem I ignored your response.

Not sure I want to swap as I think the main issue is in the hybrid chip. I'm guessing someone blew the input which damaged the hybrid, or maybe they replaced the G21 when they got different measurements like me, but replaced it with one with the sideways 32 and blew the hybrid. I'm uncertain the AD823 is bad, but this is irrelevant since it's obviously an AD823.

I'm on the fence with how I'll proceed with this scope: I can replace the components (once I can confirm what this G21 component is) and leave it with a bad 50ohm input on two channels, or gamble with the ones being sold on Ebay. Gambling is around $160 for two chips that could be counterfeit, and the cost to have someone replace them. Probably a $300 total gamble - and it's not to say the replacement chips will last since they could be counterfeit.

[lordium]

You can see picture for some samples when I check this part, they all have those numbers on the side (most likely date code or similar).

[bostonman]

Take the readings from the various pins of the transistor and evaluate the approximated values with the datasheet. So that this might help to know whether the datasheet referred is correct or not

Well technically I've done that. Between two pins, I should get 11k, but looking at my measurements, I didn't get that on any of the pins.

[lordium]

When I set up test with regular transistor (2n3904) with external resistors, and measure resistance I get the following:

~11k using UT-61E (both directions)

~11k using 3457A positive on emitter

but

~6.8k using positive on base, because the 10k is in parallel with the diode(PN junction), so the 10.24k could be due to your multimeter.

also, the 11k is not super important, since it's more likely trimmed to the ratio of the resistors, not actual value (might differ from one manufacturer to another).

[bostonman]

I think you're looking at my in circuit measurements.

If you look at my last message with the attachments, the updated spreadsheet has a section for the G21 component removed from the circuit and measured on the bench.

[bostonman]

I'm kind of at the end of the rope with this component and based some findings on the website provided above:

http://chip.tomsk.ru/chip/chipdoc.nsf/vc1!readform&view=smd&cat=G&start=501&count=500

I believe I've narrowed the possibilities to three components based on package size (see attachment that includes all the measurements of a good component and the three possibilities).

The most practical one seems to be the DTD113ZK. If it were this component, I'd assume two pins would measure 11k out of circuit. If so, seems it's back to the drawing board with figuring out what this component is.

[ataradov]

Can you actually measure voltages on the pins when the board is powered on.

If your schematic is correct, it makes no sense to have a transistor there.

Also, how does the channel behave when the part is not in circuit? It is not breaking the signal path, so you should see something.

[bostonman]

I was hoping to solve this before powering the board as I've removed a few components for troubleshooting purposes, however, it seems some questions will be answered if I take your approach.

My concerns are: I power it without the components, and they turn out to be something that controls a voltage thus causing a high voltage on a component and damaging something; or I install the "bad" components and need to remove them again to replace which could damage the board. I know on Ch 1 (the good channel) a pad lifted on pin one of this G21 while I was removing it.

In any case, I think the best approach will be your suggestion and just power it to confirm voltages.

[bostonman]

I'm waiting for some replacement components from Digikey. I am replacing the AD823 and resistors I removed to measure and test the circuit in both channels.

Once those arrive and are installed, I'll power it without the G21 and provide updates.

[bostonman]

Hopefully this topic hasn't gone stale.

Unfortunately I've been unable to get time to solder new components on and take measurements, but this weekend I set aside time.

Attached is the schematic (which I've previously posted) of how I interpreted the connections based on hours of tracing. Second file is the measurements of two out of the four G21 components (one per channel) measured out of circuit (note: one is G21 and the other is G21 with a sideways 32 - I suspect channel 2 and channel 3 had these replaced because these are the only two channels which are bad, have flux indicating someone replaced these, and have a '32' on them - possibly being the wrong component).

These measurements were done in both resistance and diode settings as noted.

Third file are voltage measurements in all four channels (channel 1 and 4 work in 50ohm mode, and channel 2 and 3 don't and are considered "bad"). Two channels don't have the G21 installed (noted in the file), so I was able to measure the pads only. You'll see that the two G21 components removed are the ones I measured out of circuit. Since they were already removed, I kept those channels without G21.

I switched the front channels from high impedance to 50ohm to see if that affected any voltages (noted on the attached).

It seems this G21 doesn't contribute to the normal function of the circuit, but may be some sort of protection device.

My guess: these components are not the cause of channel 2 and 3 failing in 50 ohm mode, however, trying to figure out what they are so I can replace them with the correct "G21" components I think is important.

[ataradov]

All of this further confirms that this is just a diode array. It clamps the input voltage to ~1/2 of voltage rails.

For the first pass its presence should not affect performance on small scale inputs. If something does not work, then it is not the fault of that component. You can confirm that by moving the good one between the channels.

And then once you figure out what else is broken, replace it with BAT54S and don't waste any more time on that.

[bostonman]

I agree that it must be some sort of protection device (especially since it's tied to the input signal path side), however, based on the package size I measured, it only matches that of a transistor type package based off of:

http://chip.tomsk.ru/chip/chipdoc.nsf/vc1 (http://chip.tomsk.ru/chip/chipdoc.nsf
/vc1)!readform&view=smd&cat=G&start=501&count=500

[ataradov]

That list is far from complete. It also has not been updated in ages. You measured a diode, it makes sense that according to the schematic that it would be a diode. Why are you refusing to believe that this is a diode?

[bostonman]

Not refusing to believe it at all. I just wanted to remind you (and/or anyone reading) about past postings.

Much like myself, we follow different messages and/or message boards, so some messages within the thread can be forgotten and wanted to remind you that a list was provided; and that I looked through that list thoroughly checking each package size to see if it matched mine.

Since nobody commented on that list, I assumed it was accurate and what this component type would be limited to.

At this point I accept your answer as being correct. Ideally I'd like to replace this with the same manufacturer. It would also be nice if I could understand what the sideways 32 represents since it's on the "bad" channels. This could be the cause of the channels failing or maybe coincidence.

[ataradov]

Are you sure it was not a repair attempt before you? And someone just put random parts at some point. It is not clear from your pictures, but it looks like the soldering quality is different on the 32 parts.

There is no way to identify parts. The markings are mostly there for confirmation, not identification.

There is no need to replace it with the same manufacturer. It just does not make sense. This diode array does not do anything most of the time.

[bostonman]

Are you sure it was not a repair attempt before you? And someone just put random parts at some point. It is not clear from your pictures, but it looks like the soldering quality is different on the 32 parts.

You're correct, the 32 marked parts are in channel 2 and 3 (the bad channels) which have less than acceptable quality soldering. Also, there is an AD823 in both bad channels with the same soldering skills.

In hopes the Agilent hybrid chips aren't the fault, I measured random points comparing them to the good channels (1 and 4) and came across some voltage/resistance differences.

Long story short, it led me to the poorly soldered AD823. My conclusion is that someone found the same voltage differences, traced it to the same points, took a chance on the AD823 and G21 being the culprit instead of the obsolete hybrid, replaced them, didn't fix the problem, and sold the scope.

Seeing as the AD823 in both bad channels had bad soldering (and also hoping they were the culprit), I bought new ones and soldered them (in my opinion I did a much better job because I didn't have several solder balls like the other person).

At this point, I don't see any other components that were replaced because everything has good factory looking soldering. The only remaining component that has bad soldering and may have been replaced with the wrong one is the G21 with a sideways 32 whereas the good channels (1 and 4) have just G21 and nice soldering.

Sadly, after all my work, I need to make a decision on whether to gamble with buying hybrid chips. The only ones I found are on Ebay from China and I've been told to watch for counterfeit parts. After buying them, I may need to find someone who is better skilled than I at replacing them.

Currently all the channels work in high Z input mode, but I can't calibrate it because it aborts when it starts channel 2. I assume because I can't calibrate it, this explains why each channel has a different offset. Channel 1 and 4 work fine in 50ohm mode, but the signal collapses on channel 2 and 3 in 50ohm mode.

I'm fearing that I buy hybrid chips, pay to have them replaced, and channel 2 and 3 are complete useless because they are counterfeit (or just junk). At that point I'd need to pay to have the old ones resoldered onto the board. Unfortunately I haven't found any information on the hybrid pin signals to take measurements.

[ataradov]

Can you trace any controls for the 50 Ohm mode? If it is all internal to the hybrid, then the answer is obvious.

[bostonman]

50ohm mode is driving me crazy.

I have yet to see a 50ohm resistor (or any unmarked resistors that measure 50ohm). Also, from what I can tell, if you look at the photo I took of the G21 component in the circuit, you'll see the single pin side is connected to a capacitor that goes into the hybrid. I'm guessing this is the main path into the hybrid, and all the relays before control whether to connect 50ohms to ground or not.

When I measure this point in high Z, I get a signal, when I measure it in 50ohm mode, it vanishes; and I can't locate a relatively normal looking waveform at any point.

The only POSSIBLE component (other than the hybrid) that could create 50 ohms is an IC on the backside labeled V224NS with (what I assume) is the company name above it: NAIS. The datasheet (I found a Panasonic one) describes this as having resistance, but, for the part number, it's less than 50ohms whereas I'm measuring almost exactly 50ohms.

A thought on the HP discussion board is: because I can physically measure 50ohms at the BNC connector on the front panel, then maybe a calibration will solve the issue of not working in 50ohm mode and it's not the hybrid. If I attempt a calibration, channel 1 (the good channel) goes smoothly and completes, I get an "Acquisition failure in cable connections test" error when starting channel 2, it starts after clicking 'OK', it continues for about ten-seconds, Windows error about the Agilent AG5483x software aborting, the program aborts, and I need to reset the scope.

[ataradov]

50 Ohm termination is likely to be inside the hybrid.

At the same time, your relay placement makes no sense. Are you sue it is not across the capacitor to make AC/DC work?

Also, what is the source of your test waveform? Are you sure it is not just loaded with 50 Ohms and can't drive a load that high?

[bostonman]

Also, what is the source of your test waveform? Are you sure it is not just loaded with 50 Ohms and can't drive a load that high?

That's a good question. I'm taking one of the signals from the rear of the scope (calibration out I think), which, on a good channel, looks good in 50 ohm mode (a clock pulse). I'm trying to trace this signal on a good channel just to figure out the path; again, in hopes that it leads me to something other than the hybrid. Sadly the signal (as I mentioned) seems to vanish, or get so small, in 50ohm mode, it doesn't make sense that it would decrease enough that noise would be injected onto it as it travels through the circuit(s).

As you can see from the pictures, several relays exist on the input side, and the V224NS is on the back side. I included a picture with red circles. The two resistors on the right in ch3 and ch4 were replaced by me. I found a difference in voltage on these and hoped the resistors had open end caps. I needed the resistors in the "good" channel (ch4) to measure the unmarked resistor values; and why a good channel had the resistors changed. The two top channels (3 and 4) have the G21 removed - I removed these to measure a "good" one and a "bad" on the bench - these were noted in a previous post. The two ICs in circles are the AD823 I replaced in hopes they were bad causing the 50ohm mode issue.

The schematic I drew that shows a relay on the input was noted as a "simplified section" because I'm uncertain how it's connected, however, I know relays are directing the input signal. This schematic began because I was trying to locate the difference in voltages between the good and bad channels. As you can see from the pictures, the AD823 in 2 and 3 was changed by me. The print on them is darker than the original bright white printed part numbers you see in ch 1 and 4.

[ataradov]

If the working channels work with the 50 Ohm setting and the same source, then it it not the source issue.

Another thing to do is to actually enable 50 Ohms on the channel and measure the resistance with a multimeter. Do that on the working and broken channels.

[TurboTom]

Your "ominous" SC70 component can very well be an HSMP-389C HP/Avago RF switching PIN diode. The marking is listed to be G2 and I often found that not all letters / digits present on the casing are relevant, frequently production batches, locations or datecodes (or what the facility manager had for breakfast...) are encrypted in these additional symbols. If this diode is really shot (as it appears), I've got a bad feeling regarding the hybrid...

[bostonman]

Another thing to do is to actually enable 50 Ohms on the channel and measure the resistance with a multimeter. Do that on the working and broken channels.

I've done that and all the channels measure 50ohms to ground. This was done a few months ago and why the thought of: if 50ohms is present, then maybe all it needs is a calibration and the "broken" channels will start working again, however, as mentioned above, when calibration begins on ch2 (the broken channel), the system halts and closes.

Your "ominous" SC70 component can very well be an HSMP-389C HP/Avago RF switching PIN diode. The marking is listed to be G2 and I often found that not all letters / digits present on the casing are relevant, frequently production batches, locations or datecodes (or what the facility manager had for breakfast...) are encrypted in these additional symbols. If this diode is really shot (as it appears), I've got a bad feeling regarding the hybrid...

Thanks for the feedback.... and I think you're correct, the hybrids are the culprit. Going about what to do with them will be the next challenge.

[gamalot]

If you believe that it is a dual diodes in series like BAT54S, I have a photograph for you. 

https://docs.rs-online.com/d69d/0900766b813b5eb7.pdf

---

I'm sorry I saw you mentioned this device with G21 mark in your reply on another topic, I thought you haven't got the answer yet.

After I responded to you here, I noticed that TurboTom has given the correct answer.

[bostonman]

After I responded to you here, I noticed that TurboTom has given the correct answer.

He did provide an answer, and has been extremely helpful. I am quite confident this device is a BAT54S, however, I am reluctant to use this for two reasons.

Prior to the link you provided, I couldn't find a single company that produced a BAT54S with the marking of G21. Also, this is feeding (or getting fed, or both) by the Agilent hybrid chip. If for some reason this device isn't a BAT54S, and I use the wrong device, I may blow the hybrids which will destroy the scope.

The link you provided looks helpful, and it appears, according to the table on page 3, I have either a lead code 2 or C which are basically the same thing, only a different package which is a simple solution.

The only confusion: I didn't see what the '1' represents, much like I didn't see what 'H' represents on the device in your picture.

[gamalot]

The third character represents the 'date code'. It can be any letter or number or even other symbols. In fact, if you look for pictures of HSMP-3892, you will also find G2m, G2/, G2V, or G27 and so on.

[bostonman]

I saw the date code, but didn't see what '1' would represent; maybe January?

In any case, I think that PDF is extremely helpful and adds another layer of confidence to believing I can replace this with a BAT54S without damaging the hybrids (or any other components).

It does, however, deviate a bit from the initial belief (maybe from what I interpreted) that G21 is the entire marking representing the type of component.

Just to reiterate a bit on why I want to be absolutely sure and not doubting the input from many on here. This scope has a bad channel 2 and 3 (four channels total). After many hours of measuring, I narrowed down the difference in measurements between good and bad channels to an eight-pin amplifier and a G21 component, but they looked like they were already replaced (but the G21 had a sideways 32). For obvious reasons, it was just easier to replace the amplifier rather than trying to determine if it's bad; plus the soldering was poor.

My guess: a previous owner tried fixing the bad channels and discovered the same difference in measurements between good and bad channels as I. In hopes that it wasn't the hybrid (which was my hopes), they replaced the amplifier and the G21 component. Either the G21 (and amplifier) was really the culprit, and, as a result of using a G21 with a sideways 32 (which may be a totally different component) blew the hybrid. Or the hybrid was just blown and the person replaced the amplifier (that may have already been good) and a G21 but used the wrong component.

Both G21 components in the bad channels measured differently than a G21 from a good channel (both were removed from the circuit). The amplifiers were replaced with new ones (and a much better soldering job now).

Also, I gambled and purchased used hybrids. I managed to replace channel 2 and the channel seems to work. Now I've replaced the hybrid in channel 3, but, in doing so, blew off a resistor by accident with the hot air pencil (I used Kapton tape over the whole area but one component still blew off). I'm waiting on a DigiKey resistor order to replace the blown off resistor. For obvious reasons I'm reluctant to power the unit with a missing resistor to avoid damaging anything. After installing the new resistor, checking the areas again (I've already carefully measured the continuity of each pin on the hybrid to assure a good soldering connection), I'll power the unit. Assuming the channels work with the new hybrids (and my soldering job), all I'll need is to purchase some BAT54S (which I forgot to order with my resistor order).

[gamalot]

For me, it doesn't really matter what the date code is, I just need to know that it is not part of the device model.

HSMP-3892 is used in the AFE of Keysight (Agilent) DSOX2000/3000 series oscilloscopes. You can find it in some pictures in that famous very long topic:

G2d on page 39
G2/ on page 40, 42
G2Z on page 52, 56, 57
G2G on page 56

https://www.eevblog.com/forum/testgear/dsox2000-and-3000-series-licence-have-anyone-tried-to-hack-that-scope/

[TurboTom]

...

In any case, I think that PDF is extremely helpful and adds another layer of confidence to believing I can replace this with a BAT54S without damaging the hybrids (or any other components).

...

Sorry @bostonman , but I really don't understand your reasoning to install a BAT54S instead of the obviously correct HSMP-389C. Do you understand the difference between a Schottky and a PIN diode? The use of a PIN diode as a fast RF switch to short-circuit too strong signals to ground makes complete sense in this place while a Schottky diode with its 10pF+ of capacitance (in case of the BAT54S) will form a nice low pass and probably limits the scope's bandwidth to a hundred-and-something megahertz. BTW, the total capacitance of the HSMP-389 series is around 0.3pF while its RF resistance with a posistive bias applied is as low as 2.5 Ohms. That's why HPAK used this component as an actively switched RF limiter.

I've got the impression that you really didn't understand how the circuit is supposed to work. A schottky diode installed instead of the correct PIN diode may not directly harm the instrument, but it won't protect it in case of too high RF level at the input either.

You've got to be pretty self-confident to just ignore well-meant advise from other forum members that only try to help you to get your instrument going again. But maybe you just want to learn it the hard way... 

[bostonman]

I apologize….. I thought you were giving an alternate part number that was a BAT54S - like a OEM number.

I’ll read your message tomorrow (or sometime this weekend) and respond based on your input. My priority was responding now just to make sure I didn’t rub you (or others) the wrong way.

[bostonman]

I re-read the message and looked at the datasheet (again).

The mistake I made was focusing on the device being a BAT54S. When I looked at the datasheet the first time, the title of 'pin switch diode' didn't register with me.

Obviously as I looked at the configurations, looking at them being just diodes further had me focused on it being a BAT54S.

Thankfully you called me out on my ignorance thus allowing me to realize the mistake of ordering the wrong device I almost made.

It seems this device has been narrowed down to a BAT54S or the HSMP-389x, and, according to your message, I can toss out the BAT54S theory.

I looked for the HSMP, however, it's obsolete. The alternate part numbers include this one:

https://rocelec.widen.net/view/pdf/wkwnaj16jm/INFNS15694-1.pdf?t.download=true&u=5oefqw

I looked at both datasheets and the HSMP can handle 1A of forward current whereas the one in the link above can only handle 100mA. I'll look at the other alternate parts to see if the others are better.

[TurboTom]

@bostonman -
the BAR63-04W should be a perfect replacement for the HSMP-389c. Don't worry about maximum pulse current and the other DC ratings. Rather have a look at your own circuit that you posted some time ago:



Where should any substantial DC current be coming from?

C5 is only conductive for RF signals. The input of the hybrid via R3 won't supply any current worth mentioning. And the biasing networks R1/R2 and R4/R5 won't harm the diode either. The way the circuit is supposed to work is as follows: During normal operation, the hybrid biases the (double) diode to block by raising pin 18 to the supply voltage. This way, the input signal "sees" the diode only as a small capacitance to ground (some fraction of a picofarad).

If the hybrid senses the RF input signal to be strong enough to become potentially harmful, it lowers pin 18 voltage, thus causing a current to flow through the double PIN diode, reducing its RF impedance and shorting some or all of the RF to ground through C1 and C4. This way, the hybrid should be (more or less) protected.

If of course someone considers it a brilliant idea to connect the output of a multiple-watt RF PA to the input of the scope, the poor diode (and in consequence the hybrid as well) haven't got a chance and will let go of the magic smoke that made them tick... 

I hope this clarifies a little the working principle of the RF branch's input protection.

P.S.: Just noticed -- the schematic appears to be not quite accurate, the center terminal of the double diode is connected directly to pin 3 of the hybrid and the 100R resistor (R3) appears to be on the input side of the "active" circuitry. If a diode has failed catastrophically, I'ld also have a look at R3 and both caps C1 and C4. R3 should protect the diodes and also the hybrid's input fairly well from moderate overloads. If the diode got destroyed anyway, someone must have fed an RF signal into the scope strong enough to arc over a vaporized resistor...

[bostonman]

I'll look at whether I already fixed that resistor error on my schematic. Not a big deal, however, if I'm asking for help on this forum, I'll waste the time of others if I'm posting incorrectly drawn schematics. Also, it brings down my credibility if I'm posting inaccurate stuff.

As you can imagine, I spent many hours tracing just to get that small schematic. I feel lucky to get the amount traced that I did. I'll admit, the schematic didn't make much sense and questioned my tracing, but I also didn't know anything about the hybrid.

Prior to buying the scope from someone, I was told about the broken channel 2 and 3, and if a voltage exceeds a certain DC voltage (some reason 5v comes to mind) is on the input while in 50ohm mode, the channel will blow up. At that point I was deathly scared of damaging the two good channels.

Unfortunately I'll never know how channel 2 and 3 got damaged. Upon hearing about blowing up the channels due to high signals, I assumed they were sensitive and poorly designed. Seeing that the G21 in those two broken channels had a sideways 32, measured differently out of circuit, and that a G21 can vary in part type across different manufactures, I wonder if someone injected a high powered signal, blew up the original G21, replaced it with the wrong type, and blew up the hybrid as a result.

You seem to have extensive knowledge about the hybrid functions. I searched for some theory of operations, or any information on the hybrids, but never found anything.

I believe my Digikey order is arriving today. Although I used Kapton tape to hold the components while using the hot air pencil, it managed to blow off a 1k resistor when I was replacing the hybrid in channel 3. I'm uncertain if I'll install the 1k resistor today or tomorrow, but, at after that, the scope will just need the new BAR63 in channels 2-4. Although I've powered it without these installed already, maybe I'll hold off until I get the BAR63s and install them.

Do you happen to know what the other digits in the part number represent in: BAR6304E6327HTSA1

The datasheet only seems to explain the BAR6304 part, but I don't know what the E6327HTSA1 mean. Digikey has more digits after, however, it's just tape and reel, cut tape, etc...

Most likely I don't need to worry about those numbers, but thought to ask just so I can know what I'm ordering.

[gamalot]

BAR6304 is the model number

E6327 are packaging suffixes

https://forum.digikey.com/uploads/short-url/2voJtGPwjujFVCBMZiIZH5oOb3d.pdf

HTSA1 are OPN (Orderable part Number) suffixes

https://media.digikey.com/pdf/Other%20Related%20Documents/Infineon%20Other%20Doc/Part_Number_Guide_Web.pdf

[TurboTom]

I don't know much about the hybrid in particular, but just by looking at the components around and the interconnection, my educated guess would be:

- It's a summing amplifier that combines the AC and DC branches of the single-ended input circuitry.
- It contains an RF detector circuit to limit RF input level via the double PIN diode in order to protect the sensitive RF input.
- Looking at the three differential pair output lines, it probably contains individual differential Drivers for the ADCs and probably for the trigger circuitry.

Once again, I'm not in any way an expert for these scopes, I even haven't got a HPAK scope, but with a little experience it's not that difficult to get an idea of what's going on in there.

I recommend to get exactly this diode: https://www.digikey.com/en/products/detail/infineon-technologies/BAR6304WH6327XTSA1/12686013

The "W" type is the one you want, it's an SOT-323 enclosure that's smaller than standard SOT-23 and it will exactly fit the footprint  on the PCB. Moreover, the smaller casing has lower parasitic capacitances / inductances. Don't get the "E" version!

[gamalot]

I don't know much about the hybrid in particular, but just by looking at the components around and the interconnection, my educated guess would be:

- It's a summing amplifier that combines the AC and DC branches of the single-ended input circuitry.
- It contains an RF detector circuit to limit RF input level via the double PIN diode in order to protect the sensitive RF input.
- Looking at the three differential pair output lines, it probably contains individual differential Drivers for the ADCs and probably for the trigger circuitry.

Once again, I'm not in any way an expert for these scopes, I even haven't got a HPAK scope, but with a little experience it's not that difficult to get an idea of what's going on in there.

I recommend to get exactly this diode: https://www.digikey.com/en/products/detail/infineon-technologies/BAR6304WH6327XTSA1/12686013

The "W" type is the one you want, it's an SOT-323 enclosure that's smaller than standard SOT-23 and it will exactly fit the footprint  on the PCB. Moreover, the smaller casing has lower parasitic capacitances / inductances. Don't get the "E" version!

I have not noticed their size, yes you are right, they should be SOT323 package. If the OP doesn’t mind buying components from China and knows how to buy them on Taobao, HSMP-389C is still available.

[bostonman]

Buying from China probably isn't a good idea since ideal substitutions exist; unless others think differently.

Unfortunately I no choice but to purchase the hybrids from China. They came soldered to a PCB that was cut from a larger board. I'm guessing (and hoping), the person or company cuts them from old HP equipment and they are truly Agilent chips and not knock offs.

I'll know within a week or so if those hybrids are good as I'll order these BAR components tonight and have them soldered when they arrive. At that point I can then power the scope and run some tests.

As for the SO23 versus SO323, looking at my case dimensions (see attached), neither of these are the same size (assuming I measured my component accurately), and the SO23 looked closer to the actual size. This doesn't mean the PCB footprint wouldn't support the SO323 or I couldn't make it fit on the existing footprint.

As it was pointed out, seems electrical advantages exist using the smaller package as parasitic capacitance/inductance is lower, so I should try making the SO323 work regardless.

[TurboTom]

What about sending one of the broken hybrids - after you pulled them from the scope - to @Noopy, our resident decapping king? He's a real wizard at taking superb quality micro photographs of the internals and this way, some knowledge could possibly be gained about the way these circuits work in detail! By doing so, you could return some of the favors you got in forms of consultation back to the forum  .

Btw, the components in the shape of an SOT-23 are only available in two sizes, SOT-23 (small) and SOT-323 (very small). If you look at your own photo of the concerning area of the PCB, you will find a semiconductor in a SOT-23 casing right below the "NAIS" relay. And that appears big in contrast to the double PIN diode, which, consequently has to be in an SOT-323 enclosure. Don't worry about taking measurements, the "W" version of the BAR6304 diode will fit the footprint on the PCB perfectly well.

[bostonman]

That's interesting, I didn't know someone on here decaps chips.

It's a good idea, however, these chips work for the most part, just (maybe) not in 50ohm mode; more when I get the scope tested. Either way, I'll keep this in mind and contact him should I want to sacrifice a chip.

I just placed another Digikey order, this time for 15 of the BAR64W. Although I don't have any intentions of using these, I also bought 15 of the E version and 10 of the BAT54S.

One thing about Digikey: you pay $4.99 shipping costs for a box the size of an encyclopedia, and it contains enough components that can balance on the head of a pen. I might as well just dump a few extra components on the order just for my inventory.

[comka]

Hi guys  !

can to whom It will be useful, there is a good frame from my scope 
It looks like an earlier printed circuit board assembly,
This can be seen According to the production of components,
as well as on a hybrid assembly of dated 2006 years.
It can also help see what year And what kind of plant
And how the component marking is applied.
This is useful to identify the original components.
Successes in repair.

[bostonman]

Funny enough, that component isn't marked G21, it's marked G2 with a sideways 11.

[bostonman]

I received the replacement components. Unfortunately I've been too busy to solder them, however, tonight I measured them. Although I was quite sure they'd measure similar, I thought double checking wouldn't hurt.

Attached is a comparison of the G21 component, the one with the sideways 32 (either it was blown or someone chose the wrong type because the measurements are different), and the BAR6304. Also, just to show which component I was measuring, I included a picture of the bag.

I'm anxious to get these soldered in, and hopefully will have time Friday or over the weekend.

Edit: I plan to contact Noopy about sending a chip to him to slice. I need to debate whether I should sacrifice a working chip to some degree in order to ruin. It seems reasonable to do in order to gain knowledge about the chip.

[TurboTom]

The replacement diode compares completely fine to the good "G2" of CH4. Differences of the threshold voltage by a few tens of millivolts are normal and the result of slightly different doping processes (different age / fab). Remeber, this diode is supposed to be operating in its "blocking" region during normal operation of the scope. Slight differences in the voltage drop in the conduction region are irrelevant, important is the thickness of the "intrinsic" zone and thus the lifetime of the charge carriers within it. This is what causes the RF conduction of a PIN diode when forward-biased.

[bostonman]

I assumed the differences were perfectly fine, but wanted to post them since many have taken an interest in this component. I was quite pleased with the readings I got, but somewhat expected them to match since the diode configuration made perfect since once light was shed on the device.

I'm anxious to get these installed and test the scope; and nervous too. Obviously I'll post my results once I get the scope running and tested.

Looking at that sideways 32 component and the difference in out of circuit pin measurements, I wonder whether this component is blown (maybe someone replaced it, in turn it damaged the hybrid, a new sideways 32 chip was installed, and the damaged hybrid blew the component) or not blown and just the wrong component.

Unfortunately we'll never know.

Any guess as to how much this would clamp a high input voltage before the hybrid got damaged? As I mentioned, upon buying this scope, I was told that anything over 5V DC while in 50ohm mode will blow the input channel. I find this hard to believe.

[bostonman]

I soldered the BAR6304 components and today ran a self-calibration.

I'm happy to report all four channels passed calibration. Whether the pin diodes will protect over voltage on the input is another story, but I appreciate all the help. As with a few other threads I started, this has been one helpful thread and I'm glad to see such a message board exists to share knowledge.

I'm uncertain whether I should ask this in a new thread, but, when I went to re-seat the RF shields, I noticed one channel has a different series capacitor from the BNC to the circuitry (see attached).

Unfortunately I doubt finding the value will be easy since I don't have smart tweezers, and I'm uncertain if my Fluke will measure such a capacitor (I'm guessing it's in the Pf range). Obviously I need to remove a good capacitor to measure the value, and I'm a bit scared to tinker with what works.

If I can't solve this capacitor issue, I may just leave it as is. The scope seems to work.

[Zero999]

I wouldn't worry about the capacitor being different, but the soldering is dreadful. I'd remove it and resolder it properly.

[bostonman]

I wouldn't worry about the capacitor being different, but the soldering is dreadful.

In all cases where a component was replaced, this was a typical soldering job. I'm not great at soldering surface mount, mainly due to using a bit too much solder, however, I make it a point to center them and make sure they are flat. Not to mention, cleaning so everything looks shiny. I'll assume the person who replaced the components didn't have a good microscope.

The eight-pin op-amp or whatever it was had solder balls. They were big enough and close enough to pins that I'm shocked it didn't short (or maybe it did and blew the hybrids). Thankfully those were easy enough to replace and make them look new again.

Speaking of centering components and making sure the soldering is good, I totally forgot, the footprint for the SOT-323 fit perfectly.

Ideally I'd like to replace the capacitor. I've really cleaned this scope and leaving a wrong capacitor would be a sin at this point, but it will probably be too difficult to figure out the value since caps have a high tolerance. Due to the soldering job, and the person using a G21 with a sideways 32 (which measured differently), one has to wonder how they chose this capacitor value.

For all I know, it could be the wrong value and causing the input signal to lose amplitude due to impedance mismatch.

Edit: does a more accurate way exist to measure the value of the component? I have a Fluke 179, and I believe I have other meters too. I think I'll go for the big prize and remove a good capacitor to measure (and measure physical size).

[bostonman]

I didn't remove a good capacitor to measure its value, but I began by looking at how it's connected. Fortunately it's connected across a V224NS which is a FET relay.

This capacitor is in series with two 1.96k resistors (also in series) which go across pin 4 and 6 (but they also branch out to other circuitry).

I was hoping looking at the datasheet would specify the capacitor to use, especially since this is a CR snubber, however, I didn't see a value.

The cap size is approximately 0.1" x 0.124".

I plan to remove a good one and measure the value.

Does that look like a ceramic capacitor (so I know what type to search for)?

[Zero999]

If it's working, then both capacitors will be good.

I'd leave well alone, unless it's to fix the crappy soldering.

[gamalot]

For MLCC, especially those with relatively large size, I will choose to replace it with a new one or leave it there without touching it. Re-soldering may run the risk of damaging it due to uneven heating.

[bostonman]

For MLCC

I assume Multi Layer Ceramic Capacitor?

I am 90% in favor of leaving this (unless it's to fix the solder). I'm assuming the difference in physical size is due to under voltage value that may short at some point.

[bostonman]

For MLCC, especially those with relatively large size, I will choose to replace it

Out of curiosity, but how did you know this is a multi-layer capacitor? Or maybe the better question: what physical properties does this have that determines it's a MLCC?

[gamalot]

I haven't seriously studied what the precise definition of MLCC is, and my English has restricted me from describing it in detail. My thinking is simple. If it looks like the MLCCs I have seen in the past, then it should be an MLCC. Although their colors will be different because of the use of different materials (X7R, X5R, COG, Y5V, etc.), it is still easy to put them into one category.

[bostonman]

Stating it's a MLCC is a big help because I thought due to it's size, it was just a higher voltage rated capacitor, however, now I know it's a MLCC.

Because it's a snubber circuit, I want to be sure it protects the component(s) it's designed to protect, and not keep the wrong type/size/value in the circuit.

I'm fearing if I remove a good capacitor to measure the value, it may be out of range for my Fluke. Plus, I still need to figure out if it's a X7R, YSV, etc...

It would be nice to return to the days when components were big and had markings.

[bostonman]

You seem to be correct about it being a MLCC because I removed a good capacitor to measure, and got 24pF.

When I start the filtering process on Mouser (and maybe DigiKey), I begin with 24pF and a 1210 body (I measured the capacitor size to be 0.1" x 0.12) since those are the two things I'm fairly sure of and it basically shows me just MLCC.

I'm going to measure the capacitor with another meter to see if I also get 24pF because I'm finding only around 27pF (.027uF) values I think.

Edit:

I was re-reading a few messages back and realized I made an error when I typed:

I just placed another Digikey order, this time for 15 of the BAR64W. Although I don't have any intentions of using these, I also bought 15 of the E version and 10 of the BAT54S.

Obviously I meant BAR63W, however, it's obvious I ordered the correct ones because I included a picture of the bag in a previous post.

[Zero999]

Are you sure it's a snubber circuit? 24pF is a 5% tolerance value, or better, which is quite close for a capacitor. Snubbers generally don't require close tolerance parts, 20% is often adequate. If it really is a snubber, 5% tolerance might be used to simplifiy the BoM: it's easier to use all 24pF 5% capacitors, rather than a few 24pF 5% and a couple 22pF 20%.

[TurboTom]

This cap appears to be the input AC coupling (DC blocking) element since it's basically in parallel to a photomos relay (this black NAIS labeled component). As such, I'ld expect it to have a rather large(ish) capacitance, at least 100nF, maybe 220 or 470. The 1210 form factor makes me assume its voltage rating is at least 250V, maybe even 400V (since that's the rating of the photomos relay, I assume rather the latter is correct, possibly even higher).

The way to appraoch unknown components is to look at the surrounding circuitry, understand what they are supposed to do and then decide if a measured figure makes sense in this environment. 24pF definitely doesn't seem right to me for this cap.

[bostonman]

The assumption of it being a snubber is due to page 7 of that datasheet which mentioned it being a snubber. Obviously I'm uncertain and open to accepting it being used differently.

I fully agree with you on the voltage. I planned to buy the largest voltage rated capacitor I could; unless of course it became so big that it wouldn't fit under the RF shield.

I already supplied the measurements, but tonight I measured the height to be approximately: 0.065" or 1.65mm

Attached is the measurements using different meters. To make sure I got good connections, I soldered fly leads to the capacitor even after I measured it without the leads and got approximately 24nF.

The first meter I used (not shown) was a Fluke 179, and this is where I got 24nF. Initially I used the test leads connected to clip leads, thus making the leads far too long. Unfortunately this was the only way to make a solid connection, but eventually I used just the clip leads and held them inside the meter test lead holes (although I didn't see much difference in capacitance using shorter leads).

After I used another meter, and, to make sure meter lead length wasn't an issue, I held the capacitor directly to copper wire I shoved in the holes due to them not accepting the larger size plugs on my test leads. The clip lead holding the fly lead on the capacitor isn't connected to anything and just used to prevent my body from altering the capacitance measurement by holding it directly.

You'll see in all cases the capacitor measures approximately 23nF, however, I'm still open to the possibility this capacitor should be a higher value.

The replacement capacitor someone shoved in another channel/circuit is a typical 0603 (?), so I assume it's not the correct voltage. Since I didn't see any capacitors in the other channel with poor soldering, I'm confident someone didn't take time to remove a good capacitor and measure it like I did. My point is: it's probably good I'm taking time to replace the small 0603 capacitor.

On a side note, I sent an email to Panasonic about whether a spec exists for that capacitor value in hopes this is a external value called out for their part. So far I got the request on providing my company name, predicted volume, whether it's a new project, etc... After stating that I'm just a hobbyist requesting technical information, it was asked I provide expected voltage, load current, etc...

Since I don't know any of these numbers, I guess I'm out of luck in taking that route.

[TurboTom]

22nF or 27nF may be correct. Before, we talked about picofarads, but as I now see this may just have been the result of an error converting from microfarads...

Check this search page on Arrow for a selection of 27nF 500V 1210 MLCCs that would surely do the job. I decided to rather chose 27nF since MLCCs from experience usually measure below their rated capacitance, so 22nF may be too small. Anyway, this is not the place for a precision component, and some 10~20% too high wouldn't be an issue.

[bostonman]

Before, we talked about picofarads, but as I now see this may just have been the result of an error converting from microfarads...

I'm mad at myself. I re-read my older message and realized I kept typing pF instead of nF.

I made the same mistake in my last post, however, I caught myself just before sending it. Typically I've only dealt with uF and pF only, and I'm not familiar with having to type nF values.

Making these mistakes aggravates me because I waste my time, and the time of others.

[bostonman]

I think this Acquisition Board for the scope is at the completed stage.

The recent findings was a broken ZIF connector and the capacitor in question (although I knew about the ZIF connector sometime ago).

Anyway, I've gone ahead and replaced the ZIF connector and capacitor. As suggested, I should go with a 27nF, 500V. After digging in DigiKey, I found a 630V and a 22nF capacitor 1000V. I ordered both in hopes the 1000V 22nF would be a bit on the high side (closer to 24nF), however, both capacitors measured near their exact value.

Unfortunately the 1000V seemed a bit too tall. Although it looked like the RF shield would have enough clearance, it was very close. It turns out the 630V 22nF looked like the exact height of a good one. This alludes to the comforting feeling that 630V is either the correct voltage rating, or at least 130V above the original (i.e. 500V).

Upon removing the obvious undersized capacitor and measuring it, I believe (I'll need to double check) it was up around 330nF.

At this point I've replaced all the components in question that were different than those in a good channel; including the two bad hybrid chips.

If not for this message thread, I doubt this scope would have been repaired. Thanks for all the help.

The only remaining change is an upgrade on the power supply fan (arriving tomorrow). This will make the scope run quieter.

After the fan and reassembling the scope, it should pass calibration again since it passed before, and also since nothing major has been changed (except the capacitor and the ZIF connector which is for the Probe Board ribbon cable).

Thanks again

Edit: the capacitor I used is 478-12869-1-ND (the 1000V one that was too large and not used is: 399-C1210C223JDGACAUTOCT-ND)