Banggood 50 ohm BNC feed-through terminator - a quick review

[TheSteve]

Some time last year someone posted that Banggood had 50 ohm feed-through terminators for what is a reasonable price(6.66 USD shipped for a pair). I ordered a pair in December last year but they never arrived and a few months back Banggood issued me a refund. I was shocked yesterday when they arrived. It only took 7 months...

I did a quick check with a DMM and found one was near 50 ohms, the other though is 100 ohms - so quality control may be an issue.
Long story short the 50 ohm one actually works reasonably well.



Here is a picture showing SWR(return loss). I wouldn't really want to use these at 1 GHz but that shouldn't ever happen as any scope with such a high frequency will have a builtin 50 ohm terminator. For those with scopes up to 500 MHz without builtin 50 ohm terminators these will do the job though.


Resistor is close enough:

[BravoV]

For those with scopes up to 500 MHz without builtin 50 ohm terminators these will do the job though.

Steve, thank you, this is exactly what I'm expecting to hear. 

[alm]

The 100 Ohm one might be an open resistor if they used 2 100 Ohm chip resistors in parallel.

That seems very close to the specs of the Pomona 4119-50: VSWR < 1.1 up to 250 MHz and < 1.2 up to 500 MHz. That one uses a carbon film resistor. The Tek 011-0049-01 is better (VSWR < 1.2 DC-1 GHz). But still sounds like excellent value for the money.

[BravoV]

The 100 Ohm one might be an open resistor if they used 2 100 Ohm chip resistors in parallel.

That seems very close to the specs of the Pomona 4119-50: VSWR < 1.1 up to 250 MHz and < 1.2 up to 500 MHz. That one uses a carbon film resistor. The Tek 011-0049-01 is better (VSWR < 1.2 DC-1 GHz). But still sounds like excellent value for the money.

Yep, value for money for < 500 MHz works its very good.

As its full metal body, the physical built and mechanically is pretty decent imo, took a close up shot -> HERE (Beware of huge photo)

[TheSteve]

So far the 100 ohm one doesn't want to come apart. I haven't gone all crazy on it though.

[bjcuizon]

How about making one yourself, like mine..
https://www.eevblog.com/forum/projects/home-made-feed-through-terminator-for-bnc/msg1237711/#msg1237711

[bjcuizon]

By the way, can you give me a ballpark figure of what the VSWR is at 300MHz for this DIY Terminator I posted above? Given that I soldered the resistors on opposite sides to minimize stray capacitance. I cannot measure its performance because I do not have the right kit to do it. Thanks.

[TheSteve]

I can't really guess what it would be, but it is likely pretty decent. I know on my home brew SMA loads using three 0.1% 150 ohm 0603 resistors gives the best performance.

[BravoV]

So far the 100 ohm one doesn't want to come apart. I haven't gone all crazy on it though.

I tried too, looks like they glued the metal parts inside, still can't find way to open it up without destroying or at least chipping out the metal parts. 

[TheSteve]

So far the 100 ohm one doesn't want to come apart. I haven't gone all crazy on it though.

I tried too, looks like they glued the metal parts inside, still can't find way to open it up without destroying or at least chipping out the metal parts. 

OK, wrestled it open(they are threadlocked) - a pair of 100 ohm resistors.

[bjcuizon]

That's pretty neat though. What are the sizes/wattage of the resistors?

I also wonder why tek/hp feedthru terminators can cost much. Are we paying for the name brand it has? Or does it have really really high quality parts used?

[TheSteve]

They look to be 1206 resistors. I resoldered the resistors on the bad one and it is 49.75 ohms now. I am not convinced it will keep working though. While the resistors didn't fall off I can't help but think there is a crack in one of them.
The name brand ones are overpriced but I would think the internal construction is probably a decent amount nicer. I doubt they have hand soldered smt resistors going to the center conductor. Right now any stress on the center conductor can break the resistors.

[BravoV]

Steve, if you decided to fix it with the resistors like the ones you put in your home brew load as you mentioned above, please update us.

[lukier]

Thanks for testing. I just got 4 of these (I guess the same) few days ago:
https://www.aliexpress.com/item/P7001-2PCS-High-quality-BNC-to-BNC-Female-seat-50KY-50-ohm-matching-device-Q9-adapter/32661228102.html
and I wondered how good/bad these are. Thanks to your teardown now I see these are pretty much the same as the DIY ones I made a while ago

[BravoV]

Thanks to your teardown now I see these are pretty much the same as the DIY ones I made a while ago

As its clear it can be opened, I guess its time to mod it with better resistors or even construction to make it better at high freq ?

[lukier]

As its clear it can be opened, I guess its time to mod it with better resistors or even construction to make it better at high freq ?

Personally, I don't care as I only use such terminators on low end scopes that lack 50 Ohm input (Rigol DS1054Z, Micsig TO1074) and my higher performance scopes (LeCroy WaveRunner 6100A, Tektronix 2467) have 50 Ohm pathway, as IMHO every scope > 100-200 MHz should do.

If you are in the RF world then it might matter, but to be honest BNC connectors are not the best for higher frequencies anyway.

[tggzzz]

I also wonder why tek/hp feedthru terminators can cost much. Are we paying for the name brand it has? Or does it have really really high quality parts used?

You are paying to avoid wasting your valuable time, by avoiding "surprises". The surprise in the first message is easy to spot, but you can waste a lot of time/effort with more subtle and/or intermittent faults.

"No DP manager lost their job because they bought IBM", and "no EE lost their job because they bought HP/Agilent/Keysight/Tek"

[Pinkus]

I also wonder why tek/hp feedthru terminators can cost much. Are we paying for the name brand it has? Or does it have really really high quality parts used?

You are paying to avoid wasting your valuable time, by avoiding "surprises". The surprise in the first message is easy to spot, but you can waste a lot of time/effort with more subtle and/or intermittent faults.

"No DP manager lost their job because they bought IBM", and "no EE lost their job because they bought HP/Agilent/Keysight/Tek"

I wonder how a feed through terminator from a high brand will be constructed.
If you look at this picture
https://www.eevblog.com/forum/testgear/banggood-50-ohm-bnc-feed-through-terminator-a-quick-review/?action=dlattach;attach=329772;image
it is easy to see: you just have to accidentally drop it once (maybe it already was during shipping) and you can never be sure if each of the fragile resistors will have survived this, or if -due to the mechanical shock- there are micro cracks somewhere which will result in unpredictable results ... now or in a year from now.
Are Fluke, Tektronix or Keysight feed-through terminators using the same construction? Any volunteer with a jab saw?
Though I remember that a Hameg (=Rohde & Schwarz) terminator was not much different  (https://www.mikrocontroller.net/attachment/98198/HZ22.jpg)

[CJay]

I wonder how a feed through terminator from a high brand will be constructed.
If you look at this picture
https://www.eevblog.com/forum/testgear/banggood-50-ohm-bnc-feed-through-terminator-a-quick-review/?action=dlattach;attach=329772;image
it is easy to see: you just have to accidentally drop it once (maybe it already was during shipping) and you can never be sure if each of the fragile resistors will have survived this, or if -due to the mechanical shock- there are micro cracks somewhere which will result in unpredictable results ... now or in a year from now.
Are Fluke, Tektronix or Keysight feed-through terminators using the same construction? Any volunteer with a jab saw?
Though I remember that a Hameg (=Rohde & Schwarz) terminator was not much different  (https://www.mikrocontroller.net/attachment/98198/HZ22.jpg)

I reckon that as neat as the Banggood ones are, they will suffer from just being used let alone dropped, every time it's inserted into a 'scope or has a plug inserted it'll place mechanical stress on the solder joints unless it's potted somehow and I've no idea how that might affect the frequency response.

The Hameg one looks a better solution though I'd be surprised if the higher end, higher frequency ones from Fluke, Tek or Keysight are constructed that way, I'd suspect they're using something like the method used in the step attenuator here:

[stj]

how would this compare to using a T-piece and a 50ohm terminator originally intended for networking??

[Someone]

I also wonder why tek/hp feedthru terminators can cost much. Are we paying for the name brand it has? Or does it have really really high quality parts used?

You are paying to avoid wasting your valuable time, by avoiding "surprises". The surprise in the first message is easy to spot, but you can waste a lot of time/effort with more subtle and/or intermittent faults.

"No DP manager lost their job because they bought IBM", and "no EE lost their job because they bought HP/Agilent/Keysight/Tek"

I wonder how a feed through terminator from a high brand will be constructed.
If you look at this picture
https://www.eevblog.com/forum/testgear/banggood-50-ohm-bnc-feed-through-terminator-a-quick-review/?action=dlattach;attach=329772;image
it is easy to see: you just have to accidentally drop it once (maybe it already was during shipping) and you can never be sure if each of the fragile resistors will have survived this, or if -due to the mechanical shock- there are micro cracks somewhere which will result in unpredictable results ... now or in a year from now.
Are Fluke, Tektronix or Keysight feed-through terminators using the same construction? Any volunteer with a jab saw?
Though I remember that a Hameg (=Rohde & Schwarz) terminator was not much different  (https://www.mikrocontroller.net/attachment/98198/HZ22.jpg)

You could easily damage a brand name attenuator or terminator by dropping it, the extra cost is partly in the calibration/verification and partly the nice box!

[CJay]

who would this compare to using a T-piece and a 50ohm terminator originally intended for networking??

In the past I've found that arrangement OK up to around ~100MHz but that was the limit of what I could measure at the time, I've not checked them again since I got better gear but I'd expect the lead length to have some effect.

They're usually 'just' an axial leaded resistor crimped into a BNC centre pin and the trailing end is usually held in contact with the shell by a tight fitting plastic cap, there are some (DEC) which have a metal cap crimped over the end which may be better

[djnz]

I have bought these from Aliexpress in the past. Same problem with mine - one was 50 ohm, the other was 100 ohm. I told the seller about this and they sent me two more 50 ohm ones, which turned out to actually be 50 ohm.

[idpromnut]

I have built a couple of 50ohm terminators using 2 vertically mounted SMA connectors (of opposite gender): soldered the four posts back to back, and then bridge the conductor pins with 2 100ohm resistors. Shield the whole mess with some copper tape and you're done.

[TheSteve]

how would this compare to using a T-piece and a 50ohm terminator originally intended for networking??

As was already mentioned those terminators generally perform pretty terrible over 50 MHz due to the internal design. If I have time this evening I'll plot the SWR on one.

[stj]

i actually thought if anything, the T-piece would be the problem by causing reflections.

[alm]

Not if the terminator has terrible VSWR above 50 MHz (I have not verified it, but 10baseT does not require it, and I trust TheSteve's assertion). The short stub from the terminator (assuming you connect it directly to a high-impedance input) will not become an issue until much higher frequencies.

Has anyone observed a difference between the cheap 10base2 tees and proper RF ones? I would expect the construction to be quite similar, but I have never used the 10base2 ones for critical work.

[edpalmer42]

I've bought various terminators over the years.  Most of my work is at 10 MHz or less so just about anything will work.  But I've measured some terminators that are likely 10base2 at higher frequencies so I'd know which ones to grab if I needed to.  At 100 MHz, they all had a return loss > 30 dB (VSWR < 1.06).  At 1 GHz, they varied from < 8 dB (>2.3) to > 18 dB (< 1.3).  I've also measured various unmarked SMA terminators.  They all measured > 28 dB @ 1 GHz.  To make the measurements, I used a signal generator, directional coupler, and RF power meter.  Calibration is questionable so take the actual numbers with a grain of salt.

I suspect that if you tore apart a name-brand multi-GHz rated terminator, you'd find a very special style of resistor.  Maybe a circular resistor that would be comparable to an infinite number of resistors in parallel instead of two or three resistors.  Continuous soldered ground around the outside, no discontinuities anywhere, etc.

Bottom line, if you can measure your terminators, or you work at low frequencies, 10Base2 might be all you need.

Ed

[TheSteve]

This is a standard low cost BNC 50 ohm terminator - nothing special, it likely represents what you can expect from one.
Good enough to use to 200 MHz +

[CJay]

This is a standard low cost BNC 50 ohm terminator - nothing special, it likely represents what you can expect from one.
Good enough to use to 200 MHz +

That's nice to see, it's satisfying to see my simplistic measurement was on the money.

Have you a 10B2 T-Piece and would you mind throwing it into the mix to see how that performs?

[TheSteve]

I also tried it with a BNC T, it doesn't change it too much. Sorry I didn't grab a screen capture.

[bjcuizon]

I also wonder why tek/hp feedthru terminators can cost much. Are we paying for the name brand it has? Or does it have really really high quality parts used?

You are paying to avoid wasting your valuable time, by avoiding "surprises". The surprise in the first message is easy to spot, but you can waste a lot of time/effort with more subtle and/or intermittent faults.

"No DP manager lost their job because they bought IBM", and "no EE lost their job because they bought HP/Agilent/Keysight/Tek"

I wonder how a feed through terminator from a high brand will be constructed.
If you look at this picture
https://www.eevblog.com/forum/testgear/banggood-50-ohm-bnc-feed-through-terminator-a-quick-review/?action=dlattach;attach=329772;image
it is easy to see: you just have to accidentally drop it once (maybe it already was during shipping) and you can never be sure if each of the fragile resistors will have survived this, or if -due to the mechanical shock- there are micro cracks somewhere which will result in unpredictable results ... now or in a year from now.
Are Fluke, Tektronix or Keysight feed-through terminators using the same construction? Any volunteer with a jab saw?
Though I remember that a Hameg (=Rohde & Schwarz) terminator was not much different  (https://www.mikrocontroller.net/attachment/98198/HZ22.jpg)

Will tiny cracks in smd resistors cause increase/decrease of the resistor's performance over time? I have this 100R 2512 with a tiny crack which can be seen using a magnifying glass...does this matter?

[tautech]

Will tiny cracks in smd resistors cause increase/decrease of the resistor's performance over time?

With any compromise of a component's envelope you can be assured the the outcome will be terminal but you never know when.

I have this 100R 2512 with a tiny crack which can be seen using a magnifying glass...does this matter?

If you can remember exactly where it is in the years to come and the part is not mission critical, it might not matter much. If you want to forget all about it and ensure reliability.......replace it.
Sometimes this old saying serves and guides us well:
First cost = last cost.

[bjcuizon]

If you can remember exactly where it is in the years to come and the part is not mission critical, it might not matter much. If you want to forget all about it and ensure reliability.......replace it.
Sometimes this old saying serves and guides us well:
First cost = last cost.

Alright, good point right there!

Well, it was a thick film resistor. I researched about its construction and composition and the top layer is glass, so maybe the tiny crack was just on the glass/primary coating.

[tautech]

If you can remember exactly where it is in the years to come and the part is not mission critical, it might not matter much. If you want to forget all about it and ensure reliability.......replace it.
Sometimes this old saying serves and guides us well:
First cost = last cost.

Alright, good point right there!

Well, it was a thick film resistor. I researched about its construction and composition and the top layer is glass, so maybe the tiny crack was just on the glass/primary coating.

Which has likely compromised the envelope. 

[alm]

Got a pair to have some extra around. One of them came apart while disconnecting a BNC cable. Guess they ran low on both thread locker and 100 Ohm resistors .

[Gyro]

More likely, the assembly instructions called for three 150R resistors spaced at 60' and the poor, underpaid, assembler realised that he/she could save a few seconds by stacking two of them.

[CJay]

At least they filed through the nickel or chrome to get to the brass underneath for soldering

[Gyro]

Hence the time saving - and probably the loosening too. Less burrs on the threads to make it tight.

[alm]

More likely, the assembly instructions called for three 150R resistors spaced at 60' and the poor, underpaid, assembler realised that he/she could save a few seconds by stacking two of them.

Unlikely in my opinion, given the earlier pictures that showed two 100 Ohm resistors at 180° spacing.

[rf-loop]

How it looks when terminator other end is connected to ecample 1M/18pF real scope input.  How scope internal circuit see it and how external signal see it when this spaghetti circuit say hello to coming signal. Also we need care higher frequencies even if scilloscope nameplate freq BW is 200MHz. Many times these may have real -3dB point far over 200MHz and many times -6db well over 300 or even 400. With fast risetime signals we then can see all wonderland miracles what we have seen in many places and in this forum many times.  Terminator itself may be more or less ok but is measurenment system ok if yhink 50 ohm termination. I have seen so much terrible and horrible results, also in my self made tests. Best what I have seen was in some old HP digital oscilloscope internal 50 ohm termination what can say it is 50 ohm impedance over wide freq range. Real RF guys developed and designed. Very very different what we can see in cheap oscilloscopes where these 50ohm input are just like 50ohm  DC/audio loading resistors.

But, how is feed thru terminator connected to oscilloscope input spaghetti circuit.
50 ohm terminator connected to scope input using T is more bad. Acceptable with DC and LF but also bit higher if measurement quality is first what can push to garbage collection. Still can see wonderland images.

[SeanB]

150R reel was less than half the price of the 100R reel that morning. so they used that, despite needing an extra resistor to make up the right value. Hate to think of the guy who gets the change over ones, with either resistor value being used and either 2 or 3 as well.

[BravoV]

So how can we improve this terminator say just desolder all the current resistors, and then probably replace with new and better ones, better placement of the resistors and etc ?

Or there is not so much to expect from this BNC metal bodies to make a good terminator say up to 500 MHz ? or even 1 GHz ?

Actually I've been thinking to order more and make myself various attenuators. 

[Specmaster]

I did a quick check with a DMM and found one was near 50 ohms, the other though is 100 ohms - so quality control may be an issue.

Even so, 100 ohms is way better than 1,000,000 ohms

[Diabolo]

Hello,

2 resistors 1/4W of 100 ohms in // give 1/2W of "power", 3 resistors 1/4W of 150 ohms give 3/4 W of "power" !

Diabolo

[Hydron]

I bought four and changed one to 4*200R, mostly for added power handling. Not worried about use over a few hundred MHz so parasitics aren't likely to be a problem.

Just a warning - be careful not to heat them up too much, the plastic softens easily.

If anyone decides that they want something nicer, then RS has some own-brand ones for about £10 which are silver plated and much better made (at least externally).

[bjcuizon]

Yeah, just but be careful with the RS one. I've bought one of those and the male bnc's locking mechanism went loose right away after mating it once with a bnc jack. And also, they specified the VSWR on its online datasheet to be 1.7:1 @ 300MHz ...Oh well, it should work to maybe around a couple of hundred megahertz.

[MF-jockey]

On 18.Dez.2018 I have ordered two of this feed-through terminators:
https://www.ebay.de/itm/2PCS-P7001-BNC-Device-Q9-Adapter-Feed-Through-Terminator-Wholesale/312203578748
On 28.Dez.2018 I received it, but both were broken. From center to shell both have infinitely resistance.
I told the seller and ask for good parts, but he wants to give me a refund.

[exe]

I bought similar terminators, one had 50Ohms, another one was 75. I open the latter, it was made of three 150Ohms resistos. Apparently one didn't make a contact, but visual inspection didn't find any problems at all. Soldering looked good, it's still a mystery what happened. Occasionally the broken part was measuring 50 Ohms. So, I suspect there is an internal crack or something. If/when I have time, I'll try to inspect it with a microscope (I don't remember if I did that before).

[Microcheap]

I got a pair of these feed through terminators https://www.aliexpress.com/item/P7001-2PCS-High-quality-BNC-to-BNC-Female-seat-50KY-50-ohm-matching-device-Q9-adapter/32661228102.html

It has 3x 150 ohms resistors. You may note the wrong footprints on the PCB, that's because they are reusing the PCB from another product (20:1 attenuator)

[MF-jockey]

I received the full refund without having to return the parts and then opened the parts.
There was nothing in it but a wire, see the pictures.
Speak about this:  https://www.eevblog.com/forum/testgear/banggood-50-ohm-bnc-feed-through-terminator-a-quick-review/msg2078935/#msg2078935

Now I ordered 2 of the P57 should be good enough for use with my 100 MHz oscilloscope.

[kerouanton]

Reviving this old thread !

I've been using the good old BNC-T + 50ohm terminator when feeding RF signals to my Hi-Z scope, but was wondering about changing to more convenient R&S HZ22 or Aliexpress P57 units.

Wondering if anyone has done a VSWR measurement of those "new" P57 cheap terminators up to 1GHz, as they claim it's rated.

And by the way, I'd be curious to see the same measurement for the R&S HZ22 terminators for comparison, as they also claim 1GHz too. Upon searching on Internet I found a long thread on a german forum with measurements and so, but I had to resort to deepl automatic translation. Anyway this thread looks interesting 

https://www.mikrocontroller.net/topic/204806

[TERRA Operative]

Looks like a prime candidate for a replacement PCB with a greatly optimised layout..

Something like a continuous central 50ohm microstrip or similar with 4x 200ohm 1% decent wattage resistors, two hanging off each side tied to a bottom layer ground plane.

[mawyatt]

Looks as if the PCB can be utilized as an attenuator as well as a 50 ohm thru.

Might be a good project for someone to roll a PCB and 3D printed case. Somewhere I recall that using two 100 ohms chip resistors across for each other produced a better load than 3 or more resistors. Here a 2D field solver might be useful in getting the best configuration & layout.

Best

[shapirus]

Another variant of the same "P57" thing from Aliexpress:



I have no idea how to measure VSWR, and very likely no means either, but I think a comparative estimate can be made by looking at the overshoot at the rising edge and the undershoot at the falling edge of a pulse with fast edges (presumably 300-500 ps) when the terminator is connected between a pulse generator and a 1M scope input. With this one the over/under -shoots were significantly higher than with a handmade terminator that I made from two BNC connectors of opposite genders. No screenshots, unfortunately, as I didn't make them before I broke my DIY terminator: two of the four 200-ohm SMD resistors cracked from mechanical stress, as there was nothing but them that would be holding the center conductor in place, axially, against the outer shell (and the plastic was not going a good job at it, because I overheated it too much during soldering). So I now have a useless 100 ohm terminator.

So this Chinese thingy has a nice enclosure, but a poor PCB. And an unforgivably high price of $6-7 currently. I think I'm going to make a proper board for it and then do a before/after comparison.

[shabaz]

Advance warning: apologies for adding more options to the list of possibilities!:

If it's for a 'scope that doesn't have an in-built 50-ohm termination, then a BNC T-piece can be used (attached at the 'scope end of course), along with a 50-ohm terminator. It's not great, but it should be adequate for a couple of hundred MHz (I did ask someone at Tektronix once, and they stated that was an acceptable thing to do on their MSO 2k series, which didn't have 50 ohm termination built-in). (And beyond a couple of hundred MHz BW, most 'scopes will have built-in 50 ohm termination anyway).

However, regardless, here is where a decent brand purchase makes a lot of sense (and it still won't be excellent unless you're even more selective since even branded BNC ones are often made to the operation level of old LAN terminators, I think!). On the other hand, Minicircuits, in particular, will have as decent BNC terminators as one can get for a reasonable price, e.f. BTRM-50+ is 11 euro at Mouser. That plus a T-piece would be an option anyway.

Yet another option (if a nano-VNA or similar is on the purchasing horizon one day) is to get a little SMA 50-ohm terminator, and an SMA-to-BNC adapter piece, and a BNC T-piece. That way, you only have to spend the Minicircuits-level budget on one part (the 50-ohm terminator) and you can use it with both the 'scope, as well as with typical SMA cables with the Nano VNA.

(EDIT: and if you're making your own terminator (or feed-through termination), then as @mawyatt says, two SMD 100-ohm resistors in parallel (but at 180 degree positions, i.e. not placed together) is probably the best DIY way, soldered directly on the connector, no need for a PCB unless you can keep the inductance low for the termination, unlike that really bad PCB. Mechanically it could be an issue (e.g. cracking), so you'd have to consider how you achieve such an approach, but electrically, it's a very good way. The photo below shows the 180 degree placement as an example (ignore that they are 200 ohm resistors in the photo, they were used for a different purpose, they should be two 100 ohm resistors). The epoxy was the attempt to reduce cracking risk.

[tggzzz]

If it's for a 'scope that doesn't have an in-built 50-ohm termination, then a BNC T-piece can be used (attached at the 'scope end of course), along with a 50-ohm terminator. It's not great, but it should be adequate for a couple of hundred MHz (I did ask someone at Tektronix once, and they stated that was an acceptable thing to do on their MSO 2k series, which didn't have 50 ohm termination built-in). (And beyond a couple of hundred MHz BW, most 'scopes will have built-in 50 ohm termination anyway).

If your scope has a 1Mohm 15pF input, then adding a 50ohm terminator in parallel will result in a 50ohm//15pF input. That will, of course, cause frequency-dependent reflections.

Whether that is acceptable depends on the length of the cable (which determines the frequencies) and your application (which determines what you can tolerate).

[shapirus]

So this Chinese thingy has a nice enclosure, but a poor PCB.

...and that's a single-layer board, too.

[shapirus]

I think I'm going to make a proper board for it and then do a before/after comparison.

Maybe not exactly "proper", and not the best looking, but it performs better.




Here's a waveform capture before and after. The cyan trace is a reference saved earlier, with the original PCB. The yellow trace is my current handmade one. The signal is coming from a 74LVC-based pulser. This overshoot becomes worse when the distance between the 50 ohm load and the scope input increases, e.g. if I add a length of coax cable between them. I will need to try different lengths btw -- with a 15 cm cable it is still seen as just an overshoot, and I wonder what will happen as the cable length is gradually increased.

[shapirus]

I think I'm going to make a proper board for it and then do a before/after comparison.

Maybe not exactly "proper", and not the best looking, but it performs better.

For comparison, I made a couple more feed-through terminators out of PCB-mount BNC male and female connectors with SMD resistors soldered in star arrangement on the male one (which is the closest to the scope input possible), in this case 3 x 150 Ohm.

No need to post another screenshot of the waveform to compare it to the improved P57: it is exactly the same, ideal match. In this particular case they perform equally well (or equally poorly, I really don't know how to tell one from the other). Maybe a faster scope or a VNA (which I don't have) would show some difference.

[TopQuark]

I have the Rigol ADP0150BNC, it is quite nice.

They advertise it to be good to 1GHz. Indeed it measures well at that frequency, with 1.2 VSWR at roughly 1.4GHz. Although I don't think anyone should be using an external terminator at that frequency. 

[tggzzz]

Although I don't think anyone should be using an external terminator at that frequency. 

Ideally not, but a 3dB pad/attenuator can improve the effective VSWR.

Gain is cheap; flatness isn't

[TopQuark]

Ideally not, but a 3dB pad/attenuator can improve the effective VSWR.

Gain is cheap; flatness isn't

Wise words. I measured the VSWR of various combinations of oscilloscope 1M input, oscilloscope 50R input, Rigol terminator, 3dB attenuator. The oscilloscope has 500MHz bandwidth, so realistically I'm only interested in that range.

Takeaways:
- Terminator by itself (S11 - term), and oscilloscope 50 Ohms input (S11 - scope 50) is good, not really a surprise.
- Oscilloscope with 1M input (S11 - scope 1M) is bad, as one would expect.
- Oscilloscope with 1M input, with external 50 Ohms terminator (S11 - scope 1M + term), is quite bad, as some members predicted in prior replies. The weird and wonderful 1M//16pF input channel network will cause significant mismatches. The terminator almost hurts as much as it helps at slightly higher frequencies.
- As suggested by tggzzz, placing a 3dB attenuator should add 6dB of return loss for signals looking into the scope, improving effective VSWR. This can indeed be observed with the terminator (S11 - scope 1M + term + pad) and without the terminator (S11 - scope 1M + pad).

TLDR:
- Scope built-in 50 ohm termination is the best.
- External 50 ohm in-line terminator only helps input matching at lower frequencies (< 100MHz), and it doesn't do a great job at that either. At higher frequencies even the best made terminators will not help oscilloscope input matching.
- Adding an attenuator will significantly improve input matching, probably should always be used with external 50 ohm in-line terminators.

[TopQuark]

Somewhere I recall that using two 100 ohms chip resistors across for each other produced a better load than 3 or more resistors. Here a 2D field solver might be useful in getting the best configuration & layout.

The VNA bible agrees, two 100R 0603 resistors will work better than one 50R 0603.

I hope Joel Dunsmore doesn't mind me reproducing 3 pages of his book 

[shapirus]

- As suggested by tggzzz, placing a 3dB attenuator should add 6dB of return loss for signals looking into the scope, improving effective VSWR. This can indeed be observed with the terminator (S11 - scope 1M + term + pad) and without the terminator (S11 - scope 1M + pad).

This is interesting. We place the attenuator between the terminator and the scope input, right?

A handmade two-in-one terminator+attenuator feed-through adapter should be possible to make. It will require a PCB, though, I think. Hmm. I may want to try it. I'm puzzled by those overshoot and undershoot that I'm seeing (shown in the screenshots in my previous posts) and I can't yet find an answer to what exactly is causing them. I'm curious if the added attenuator will tame them without slowing down the rise and fall times (as slowing them down intentionally is a way to reduce or remove them).

[tggzzz]

- As suggested by tggzzz, placing a 3dB attenuator should add 6dB of return loss for signals looking into the scope, improving effective VSWR. This can indeed be observed with the terminator (S11 - scope 1M + term + pad) and without the terminator (S11 - scope 1M + pad).

This is interesting. We place the attenuator between the terminator and the scope input, right?

A handmade two-in-one terminator+attenuator feed-through adapter should be possible to make. It will require a PCB, though, I think. Hmm. I may want to try it. I'm puzzled by those overshoot and undershoot that I'm seeing (shown in the screenshots in my previous posts) and I can't yet find an answer to what exactly is causing them. I'm curious if the added attenuator will tame them without slowing down the rise and fall times (as slowing them down intentionally is a way to reduce or remove them).

The attenuator/pad requires 50ohms on both sides. Therefore it should be cable->pad->terminator->scope.

The reflected signal traverses the pad twice, hence a 3dB pad will improve the VSWR by 6dB (you know what I mean!)

[shapirus]

The attenuator/pad requires 50ohms on both sides. Therefore it should be cable->pad->terminator->scope.

The reflected signal traverses the pad twice, hence a 3dB pad will improve the VSWR by 6dB (you know what I mean!)

I've just tried a 10dB feed-through attenuator (don't have a 3dB one) between the signal source (pulse generator) and the feed-through terminator. No change in the overshoot/undershoot amount. The shape of the waveform remains the same too: near perfect match (when vertical scale is set accordingly) with a previously saved reference of the same signal without an attenuator, so no change in rise/fall times either, at least nothing that can be detected with this scope (Rigol DHO800 with BW somewhere around 250..300 MHz).

Connection is like so: generator's SMA output -> 10dB SMA-to-SMA attenuator -> SMA-to-BNC adapter -> 50 Ohm BNC-to-BNC terminator -> 1M scope input.

Probably this overshoot/undershoot has nothing to do with VSWR, I have no idea at this point, and we may be talking about different things. I will create a separate topic about this, as I'm observing some curious reflection effects as well, just need to prepare and describe some test scenarios, make screenshots etc.

I do know, though, that the amount of overshoot/undershoot increases when the terminator is not made right (see e.g. the original PCB of the P57 shown in previous posts) or the distance between the terminator and the scope input is increased with e.g. a piece of coax cable or just some BNC-to-BNC adapters.

[joeqsmith]

Not sure if it helps but we talked about compensating such a terminator for a scope.  I made a short video going over the different terminators, SPICE models and my attempt to realize the compensation in physical hardware. 

https://www.eevblog.com/forum/rf-microwave/bnc-50ohm-feed-thru/

[shapirus]

Not sure if it helps

It does!

It did not give me direct answers, as it was on a somewhat different topic, but instead it gave me an idea and suggested what I was previously missing to simulate what I'm seeing on the scope screen.

The overshoot and undershoot that I'm seeing seem to be a result of the combination of the 50 Ohm shunt in the terminator and the parasitic inductance between the scope input BNC connector and the 15 pF capacitance. As this inductance is increased, the edge ringing becomes worse -- just what I observed increasing the length of the line between the terminator and the scope input.

The simulation also agrees with what I'm seeing when the terminator is not used at all and the pulse generator is plugged directly into the scope input: the ringing is significantly reduced, but the rise/fall times become slower (in reality so much so that it can be seen directly on the scope screen).

[joeqsmith]

Glad to hear it had some use. Teach a man to fish...

That video had a link to Lasmux's thread where I was showing them some data using some home made probes.  Parasitic's hurt the performance (lots of ringing) and I made one using a thick ground strap to try and mitigate the problem (like the professionals).   If you read though it, you can get a feel for what effect it had.
https://www.eevblog.com/forum/testgear/12-ghz-active-probe-project/msg5005309/#msg5005309