Need 50 ohm input on an (old) oscilloscope that has only 1M ohm inputs: BNC TEE?

[wb0gaz]

I'm in need of oscilloscope with 50 ohm input impedance on the vertical channels.

My HP 1725A has this feature, but it's got an unrelated problem I'm not wishing to fix right now, and for bonus points, I'd really like to have four vertical channels (all the same.)

I've got a Tektronix 7000 mainframe with a pair of (200 MHz) 7A26 vertical plug-ins which are 1M ohm input impedance. I'd like to find a pair of working (400 MHz) 7A24 plug-ins (50 ohm specific) but nothing seen (that's offered as working).

My working vertical bandwidth requirements are 200-400 MHz (this is a first foray into high speed digital circuitry.)

The question is this (which isn't really specific to the Tektronix 7000/7A26, but I use that as an example to ground the conversation):

Can I use a BNC TEE (male-female-female), male to the scope input, terminate one of the two female sides in a (known good) 50 ohm terminator, and use the other female side as a 50-ohm input?

If this is completely no-go idea, why?

If this is possible but with conditions, what would be the trade-offs (for sake of discussion, assume the TEE is 50 ohm BNC type and 50-ohm termination is "perfect" (no spurious reflections, no attenuation, termination has adequate power handling capacity, connectors are free of impairments, etc.) to separate the architectural question from the specifics of the devices that would provide the external 50-ohm termination.

If this is obviously easy, then why (other than convenience) would my HP 1725A (for example) have 50 ohm input impedance selection?

Happy to answer any requests for clarification or any other way to improve statement-of-problem!

Thanks very much for any thoughts (and would welcome any pointers as to where I could scare up a pair of working 7A24's)...

Dave

[The Doktor]

Get a 50Ω pass-through terminator. https://www.amazon.com/Copper-Adapter-38-5mm-Through-Terminator/dp/B07G566JC7/

[Wrenches of Death]

The now overpriced programable single channel 7A16P also supports 50 ohm and 1 Meg input. I wouldn't use more than one in a regular 7000 series scope though. Their current draw is higher than usual.

I've got a number 7704A's that I've been using since the 1990's. My preferred vertical amp setup is a single channel 7A16P and a dual channel 7A24 in each.

Watch your input voltage on the 7A24 and the 7A16P in 50 ohm mode.

WoD

[bdunham7]

Can I use a BNC TEE (male-female-female), male to the scope input, terminate one of the two female sides in a (known good) 50 ohm terminator, and use the other female side as a 50-ohm input?

If this is completely no-go idea, why?

If this is possible but with conditions, what would be the trade-offs (for sake of discussion, assume the TEE is 50 ohm BNC type and 50-ohm termination is "perfect" (no spurious reflections, no attenuation, termination has adequate power handling capacity, connectors are free of impairments, etc.) to separate the architectural question from the specifics of the devices that would provide the external 50-ohm termination.

If this is obviously easy, then why (other than convenience) would my HP 1725A (for example) have 50 ohm input impedance selection?

Yes, you can do exactly that, but unless you use very high quality parts you may be a bit disappointed, especially at 400MHz.  The reason the internal 50R impedance works better is partly because it eliminates the two connections and the short section between the terminator and the input, also with some designs the 50R input selection will reduce or eliminate (almost) your input capacitance.  You'll probably see observable attenuation at 400MHz even with good parts.

[wb0gaz]

Thanks very much - very helpful comments indeed - covers several topics of direct interest right away, so much appreciated.

I'm going to take a crack at pass-through attenuator (I'd never heard of such a device) as an initial solution, and keep open search for 7A24's (and yes, limited input voltage range is fully expected!)

[edavid]

I'm in need of oscilloscope with 50 ohm input impedance on the vertical channels.

Why?  If your fastest plugins are 7A26s, why not just use them with probes?

... for sake of discussion, assume the TEE is 50 ohm BNC type and 50-ohm termination is "perfect" (no spurious reflections, no attenuation, termination has adequate power handling capacity, connectors are free of impairments, etc.) to separate the architectural question from the specifics of the devices that would provide the external 50-ohm termination.

This doesn't really make sense.  The problem with a tee is that it inherently introduces a stub, so it can't be "perfect".  A feedthrough termination will have better SWR, unless it's really crappy.

... and would welcome any pointers as to where I could scare up a pair of working 7A24's)

Post a want ad here and/or on the TekScopes list with a reasonable offer.

[bdunham7]

I'm going to take a crack at pass-through attenuator (I'd never heard of such a device) as an initial solution,

I have the exact pass-through models linked above and at 100MHz I see ~5% (by eyeball) attenuation over using the internal 50R, at 400MHz it is about 20%.  So -0.5dB and -2.4dB, which is actually perfectly usable for the most part if you know what is going on.

[Fungus]

Can I use a BNC TEE (male-female-female), male to the scope input, terminate one of the two female sides in a (known good) 50 ohm terminator, and use the other female side as a 50-ohm input?

Yes.

[tggzzz]

I'm in need of oscilloscope with 50 ohm input impedance on the vertical channels.

If this is obviously easy, then why (other than convenience) would my HP 1725A (for example) have 50 ohm input impedance selection?

Does it? My HP1740A and Tek 2465 and most other scopes don't. OTOH unusually my Tek 485 does. The key difference is the Tek 485 has two internal attenuators, one the usual 1Mohm, the other 50ohm.

If you have a 1Mohm//20pF input and you add a 50ohm attenuator (either through or T), then you will have a 50ohm//20pF input. That will lead (pun intended) to a reflection from a 50ohm source.

Given the OP's moniker, I suspect they are a radio amateur, so they will understand the relationships between frequency, VSWR and cable length. Whether that matters to the OP is a case for them to determine.

The VSWR problem can be ameliorated in the standard fashion: use an inline attenuator so that you have source->lead->attenuator->50ohm->1Mohm//20pF. The higher the attenuation factor, the smaller the effect of the 20pF.

Here's a few quick and dirty experiments on my Tek485. For all pictures, horizontal 10Hz-400MHz 50MHz/div, yellow is a Smith chart of the reflection, blue is the return loss 10dB/div with the blue tick indicating 0dB.

The first is the Tek 485's usual 1Mohm//20pF input. The return loss is dismal: 0dB to 15dB, depending on frequency.
The second is the Tek 485's internal "real" 50ohm input. Notice the return loss is a respectable 30-40dB at all frequencies, and correspondingly the Smith chart indicates it is close to 50ohm at all frequencies
The last is equivalent to almost all scopes with an external load. It is the Tek 485 with an external 50ohm inline load and the 1Mohm//20pF input. Notice the significantly worse return loss >100MHz of -10dB, but it is respectable <10MHz.

[edavid]

Does it? My HP1740A and Tek 2465 and most other scopes don't. OTOH unusually my Tek 485 does. The key difference is the Tek 485 has two internal attenuators, one the usual 1Mohm, the other 50ohm.

Look again, the 2465 has 50 ohm termination available on channel 1 and 2.  I don't think you'll find many 300MHz scopes without 50 ohm termination.

[wb0gaz]

The continuing discussion and observations are appreciated. While I have now ordered the inline terminator devices cited early in the thread (ETA sometime late this week), the details about measuring RF reflection properties (vs. onboard 50 ohm mode) is particularly valuable. The two product options I found for BNC inline terminator are typical "no name auction site" type things, so RF properties of the device itself will be evaluated (nanovna will probably come in handy for this.)

As to why not use regular probe (was asked a few posts back) - a couple of issues drive this application - I'm seeking to characterize waveform of a CMOS-output device that also claims 50 ohm drive capability, and probe attachment is really easy if I just solder ends of coax to board points very close to the device, and really awkward (three-handed operation at least) using a conventional probe (there are currently two test points that need simultaneous examination, and in the next step, four test points.)

I'm sure there are excellent solutions for this type of probing task in a commercial context, but as was noted, I am a ham without commercial resources, so I am "making do" (the discussion here really helps making up for resource constraints!)

[bdunham7]

As to why not use regular probe (was asked a few posts back) - a couple of issues drive this application - I'm seeking to characterize waveform of a CMOS-output device that also claims 50 ohm drive capability, and probe attachment is really easy if I just solder ends of coax to board points very close to the device, and really awkward (three-handed operation at least) using a conventional probe (there are currently two test points that need simultaneous examination, and in the next step, four test points.)

If your source is 50R, then in many instances you don't need to worry too much about smallish reflections at the scope termination.  However, if your source is not 50R, as appears to be the case here, the most effective way I can think of for testing it while driving 50R would be to solder in an appropriate 50R resistor near the source and then put a good (low capacitance) x10/10M probe across that.  But go ahead with your current plans because that may work fine as well.

[tggzzz]

Does it? My HP1740A and Tek 2465 and most other scopes don't. OTOH unusually my Tek 485 does. The key difference is the Tek 485 has two internal attenuators, one the usual 1Mohm, the other 50ohm.

Look again, the 2465 has 50 ohm termination available on channel 1 and 2.  I don't think you'll find many 300MHz scopes without 50 ohm termination.

I've just measured my 2465. The return loss is, to pick a single figure:

• 485 internal attenuator: -35dB (VSWR <1.07)
• 2465 internal attenuator: -18dB (VSWR <1.28, spec 1.3)
• 485 with inline terminator: -8dB

So the 2465's internal termination is better than the 485+external terminator, but worse than the 485's "proper" attenuator.

As before, it is up to the OP to determine the significance of those for his purposes.

[tggzzz]

As to why not use regular probe (was asked a few posts back) - a couple of issues drive this application - I'm seeking to characterize waveform of a CMOS-output device that also claims 50 ohm drive capability, and probe attachment is really easy if I just solder ends of coax to board points very close to the device, and really awkward (three-handed operation at least) using a conventional probe (there are currently two test points that need simultaneous examination, and in the next step, four test points.)

If your source is 50R, then in many instances you don't need to worry too much about smallish reflections at the scope termination.  However, if your source is not 50R, as appears to be the case here, the most effective way I can think of for testing it while driving 50R would be to solder in an appropriate 50R resistor near the source and then put a good (low capacitance) x10/10M probe across that.  But go ahead with your current plans because that may work fine as well.

I'd modify that slightly, based on the observation that the usual *10 passive probe's input capacitance isn't much different to the scope's input capacitance.

Like you I would have a 50ohm termination near the source, but then I would

• make a passive resistive divider Z0 probe from a 450ohm (or 470ohm//10k) resistor and 50ohm coax. A commercial Z0 probe can have <1pF at the tip, and be useful to 1.5GHz or more.
• have a 10dB attenuator at the scope, to reduce the scope+terminator VSWR.

That overall attenuation is probably tolerable for this application.
Soldering the 450ohm+coax isn't much more difficult than soldering a coax.

As for CMOS outputs, jellybean parts can achieve this: https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg1902941/#msg1902941

[egonotto]

Hello,

I tested 50 MHz rectangle once with the built-in 50 Ohm and then with a tee with two different 50 Ohm terminations and once the tee alone with the built-in 50 Ohm.

Best Regards
egonotto

[tggzzz]

Hello,

I tested 50 MHz rectangle once with the built-in 50 Ohm and then with a tee with two different 50 Ohm terminations and once the tee alone with the built-in 50 Ohm.

Best Regards
egonotto

What's the type of the scope, its front-end bandwidth, signal risetime and cable length?

[jonpaul]

We use TEK BNC feedthru. Most of the cheap Chines knockoffs are poor quality, do not meet specs and have BNC connectors that can damage the scope BNC inputs .

Most high speed work uses a Zo probe, which can be a DIY , eg 450 Om R and thin RG 174/U 50 Ohm coax for x10.

Low Z but crazy wide band.

See Tek Ckts Concepts book on Oscilloscope probe Circuits, esp Zo probes.

https://w140.com/tekwiki/images/6/62/062-1146-00.pdf

We use P6156 TEK on 50 Ohm scopes,

https://w140.com/tekwiki/wiki/P6156

Most 1M scopes will not have the BW you seek. Bon Chance

Jon

[tggzzz]

We use TEK BNC feedthru. Most of the cheap Chines knockoffs are poor quality, do not meet soecs and have baynotes that can damage the scope connectors.

Most high speed work uses a Zo probe, which can be a DIY , eg 450 Om R and thin RG 174/U 50 Ohm coax for x10.

Low Z but crazy wide band.

Higher Z than simple 10Mohm *10 passive probes, at 100s of MHz.

But you obviously know that

[chris_leyson]

Tektronix 7A42 is a 4 channel 350MHz plugin with selectable 1MOhm or 50Ohm input impedance. Also supports triggering based on the Boolean conditions of the four inputs. https://w140.com/tekwiki/wiki/7A42

Chris

[egonotto]

...............

What's the type of the scope, its front-end bandwidth, signal risetime and cable length?

Hello,

it is a RTA4004 from Rohde & Schwarz.
1 MOhm has 0.5 GHz and 50 Ohm has 1 GHz bandwith (with risetime 350 ps).

The BNC cable length is about 1 m.

The generator rise time is about 2 ns.

Best Regards
egonotto

[bdunham7]

• 485 internal attenuator: -35dB (VSWR <1.07)
• 2465 internal attenuator: -18dB (VSWR <1.28, spec 1.3)
• 485 with inline terminator: -8dB

So out of curiosity I tried this with a NanoVNA and I got something similar, but I don't think one number can tell a complete story.  Here's the whole picture. 

In order, Tek 485, Tek 2465B, Siglent SDS2354X+.  My battery ran out before I could get a photo of the inline terminator, but it was slightly worse than the 2465B and the Smith chart trace wandered off down and right a lot more.

[tggzzz]

• 485 internal attenuator: -35dB (VSWR <1.07)
• 2465 internal attenuator: -18dB (VSWR <1.28, spec 1.3)
• 485 with inline terminator: -8dB

So out of curiosity I tried this with a NanoVNA and I got something similar, but I don't think one number can tell a complete story.  Here's the whole picture. 

Yes, and very true.

In order, Tek 485, Tek 2465B, Siglent SDS2354X+.  My battery ran out before I could get a photo of the inline terminator, but it was slightly worse than the 2465B and the Smith chart trace wandered off down and right a lot more.

Ah, someone else with good taste in scopes (And a NanoVNA!)

Your 485 trace looks noticeably worse than mine, and the S11 dip at @200MHz looks suspiciously like a cable effect. Did you do the short, open, load, though calibration with those at the end of the cable? I found the SOLT calibration made a significant difference to the results I observed.

Parenthetically, my NanoVNA's calibration has to be redone every time I turn it on or change the stimulus in any way, e.g. changing the frequency range. Most irritating.

[bdunham7]

Your 485 trace looks noticeably worse than mine, and the S11 dip at @200MHz looks suspiciously like a cable effect. Did you do the short, open, load, though calibration with those at the end of the cable? I found the SOLT calibration made a significant difference to the results I observed.

I didn't use a cable, just an SMA male-to-male and SMA-BNC adapters.  Yes, I calibrated it but that's a bit tricky because I don't really have a super precision set of standards in BNC, so I just left it open for the open CAL and then took the adapter off and used the 50R and short that came with the NanoVNA, attached to an SMA female-to-female adapter that approximated the same length.  I think it turned out OK because the open readings were completely flat and perfect, the Smith charts looked good (and very tight on the 485).  Did yours just look less lumpy or were the values significantly different?  I actually did the 485 twice as I wasn't happy with the first calibration, it changed a bit but had the same overall look.

Siglent fans will note that the SDS2354X+ was a little better than the 2465B and really wasn't that much worse than the 485 at ~350MHz.

[David Hess]

Can I use a BNC TEE (male-female-female), male to the scope input, terminate one of the two female sides in a (known good) 50 ohm terminator, and use the other female side as a 50-ohm input?

I have found that using a BNC-T with BNC termination only performs well to 10s of MHz.  The quality of BNC terminators is often questionable so maybe a better one is called for.

I have had good results using a feedthrough termination to 100s of MHz.

[jonpaul]

Tektronix BNC term feedthu are fine, vintage US made are usually OK

any cable or adapters add impedance and dégradation SWR.

All network Analyzers required a CAL before each use.

We use Leo Bodnar 40ps pulser to check Trise, BNC is on the pulser direct to scope.

BW is not the most important scope spec.

Jon