-
Importance of Good Terminators for RF / Pulse Applications
Posted by T3sl4co1l on 12 Oct, 2014 02:13 -
From previous threads, e.g.
https://www.eevblog.com/forum/testgear/50-ohm-terminators/
Here's an example of bandwidth and terminator use.
My terminators seem to be pretty awful: not sure if it's the effect of the tee, or the terminators themselves (who would use a wirewound resistor in a terminator..??!), but see for yourself.
Setup:
Avalanche pulse generator, with 20ns pulse line
10ns cable from generator to scope
At scope: tee + terminator
First, Tektronix 475, 200MHz bandwidth. Looks a little lumpy, doesn't it? The risetime is mostly there, though.
100MHz bandwidth. Still a little lumpy, so, whatever's causing those bumps, isn't really great for general purpose (Tek 465 and many other entry-level scopes are around this 100MHz baseline -- take note).
20MHz bandwidth. Not lumpy, but you can't even tell it's a pulse anymore, it's just a... pile. Note the bandwidth reduction is 2.5 times greater than the step between the first two, so this is a pretty drastic change, rather than a progression.
Tektronix TDS460, 400MHz bandwidth. You can see the wave does indeed look lumpy like that, plus some finer bumps that weren't visible before. Likely, this signal source doesn't have much energy beyond 400MHz, so that a scope of 1GHz wouldn't see much different, and would finally be observing the true risetime of this source.
External terminator disconnected (tee remains), internal terminator switched on. The change is pretty dramatic!
Tim
-
And how would that look with a high quality terminator on the "T"? Also how long is a 10ns cable? I never heard of a patch cable being referance like that. All in all it's very interesting and informative. Thanks
BillWojo -
Not sure, none of my terminators seem to be any better. It looks about the same if I shove a 1/2W carbon film resistor (minimal lead length) in the end of the BNC tee.
Probably, a good quality inline terminator, thus as close to the scope as possible (which in my case is an undefined stub length, plus 15pF at the end), would do better, but by how much over an ideal termination (in my case, the internal 50 ohm), I can't quite tell.
Tim -
There was a discussion about terminators here some months ago where I did some tests comparing feedthrough termination to internal termination to a good terminator attached to a BNC-T.
The feedthrough terminators, even the "slow" ones, that I have perform just as well as the internal termination on my 300 MHz Tektronix 2440. Tests using a good termination on a BNC-T showed noticeable not not necessarily significant degradation below 100 MHz. I think I decided that the practical limit was about 70 MHz. I did not have any Ethernet "thinnet" style terminators to try (I think they evaporated over the years.) but I suspect they would perform more poorly.
My pulse source for these tests was the less than 600 picosecond transition time from my PG506 flat level pulse generator and my 250 MHz leveled sine wave generator. I do not have anything faster yet.
-
And of course, you have to remember that equally important (or more important) than the quality of the terminator is the location of the termination and any stub, trace, etc. that exists to the measurement point (sample and hold, preamp, etc.) You can put a perfect microwave rated termination on a Tee and have a lousy result because of the stub formed by the Tee and the input to the scope's front end. This is the primary reason why internal terminations will generally outperform most external terminations, especially those connected via a Tee.
-
... You can put a perfect microwave rated termination on a Tee and have a lousy result because of the stub formed by the Tee and the input to the scope's front end. This is the primary reason why internal terminations will generally outperform most external terminations, especially those connected via a Tee.
I was just surprised that the feedthrough terminators performed identically to the 2440 internal terminations. Maybe the internal ones are not all that good. The test using the BNC-T was just for edification and returned results consistent with my expectations.
-
I compared a avanlache pulser response both thru a Tek 50ohm passthru and Lecroy internal termination. With 1ghz bandwidth the difference was pretty substantial.
Internal:
-
I compared a avanlache pulser response both thru a Tek 50ohm passthru and Lecroy internal termination. With 1ghz bandwidth the difference was pretty substantial.
Was the input bandwidth 1 GHz when the feedthrough termination was used? The fastest high impedance inputs I am familiar with are only 500 MHz which requires a 12 picofarad input capacitance. 1 GHz implies an input capacitance of 6 picofarads.
-
FWIW a couple of days ago I did some comparisons of a few probe options against an Agilent 54831D 600MHz/4GSa/s scope and an unmodified Rigol MSO1074Z-S 70MHz/1GSa/s scope. Apart from the homebrew 21:1 probe, I used stock probes supplied with each instrument, namely the Agilent 1165A 600MHz passive probe and the Rigol RP2200 150MHz passive probe respectively. Both scopes were running a single channel only at their respective full sample rates.
The source I used was a 125MHz canned oscillator that I had in stock mounted on some copper clad PCB. It is RS part number 478-5137, aka Epson Toyocom SG8002CAPCB125MHZ. It's a 3.3V CMOS output (none of your clipped sine rubbish, so it probably has an impressive signal on airband) in a 7x5mm ceramic base surface mount package.
1. The first test was using the stock 'scope probe grabber clip.
Agilent 54831D
Rigol MSO1074Z-S
2. Next, on the Agilent only, I used the ~1.5" dual lead adapter (there wasn't such an adapter supplied with the Rigol probe).
Agilent 54831D
3. Then the spring ground connection was used.
Agilent 54831D
Rigol MSO1074Z-S
4. Finally, a test with a homebrew 21:1 probe, ie a 1k resistor on an SMA connector, RG58 to scope, internal 50 ohm termination with the Agilent, BNC T piece termination for the Rigol.
Agilent 54831D
Rigol MSO1074Z-S
That final tests shows that your probably fighting a losing battle with the Rigol. However, with the Agilent, the rise and fall times are markedly improved compared to the spring clip thingy.
Cheers, Howard
-
4. Finally, a test with ... a 1k resistor on an SMA connector...
Sure doesn't look like a non-reactive type.
Other than specifically-made RF resistors, I've found that one-layer flat film SMT resistors work best. I've tried making attenuators and terminators with good-ol-fashioned Allen-Bradley carbon comp axial lead resistors but they're kinda funky above HF....RG58 to scope...
"RG58" is something that's w-i-d-e-l-y open to interpretation. I got rid of most of my "RG58" RF cables and use only RG142 and RG400 or some similar sized "6 GHz" marked stuff that came out of a mil-surplus lot. Also have some ultraflexible solid-jacket SMA-SMA jumpers, but they're only about 16 inches long.
Remember that steep edges are made up of lots of high-harmonics -- waaay up there -- so think until your giga hertz big time. -
The idea for the 21:1 probe came from here http://koti.kapsi.fi/jahonen/Electronics/DIY%201k%20probe/ where the author measures rise times <100ps with it.
I measured the RG58 cable on my HP 8753A VNA upto 3GHz, and you're right, it's not great, but this is as nothing compare to putting BNC connectors on the end, the use of T pieces, BNC terminators and interseries adapters.
Yesterday and today I did some further testing with the Rigol MSO1074Z-S only. This test uses the same 21:1 DIY probe and a specially made cable with the scope end internally terminated with an 0603 49.9 ohm resistor. This improved things, but I also noticed that reducing the sensitivity on the scope also improves the rise time significantly, I assume because it reduces the slew rate of the ADC driver opamp. It makes for a more reasonable, more "square" result. (Note that on the first screen print I accidentally had the 10x probe setting switched on the scope).
-
By the way, these two screen shots show the return loss of the modified cable with the integrated terminator, first with a 100-500MHz sweep, and secondly from 100MHz to 3GHz. The 3GHz sweep is particularly nasty, and it's why we shouldn't be using BNCs!
-
Some more test reuslts with a Jim Williams pulse generator.
Rigol MSO1074Z-S, JW pulser without coax "add on" into SMA-BNC cable with integrated 50 ohm termination at the BNC end
... add a few feet of coax onto the JW 2pF cap to extend the pulse (note that although this shows 800ps[!] it was varying between 800ps and 1.2ns)... (Edit: This test has been re-done, see about four posts newer)
... on the Agilent 58431D, using the scope's integral 50 ohm load and a 10dB SMA pad (the pulser signal was too high for the 50 ohm input).
-
... add a few feet of coax onto the JW 2pF cap to extend the pulse (note that although this shows 800ps[!] it was varying between 800ps and 1.2ns)...
What causes the variation on the Rigol?
I assume it was not seen to vary on the Agilent 58431D.
Added: Is there something very wrong with the Rigol's transient response? Shouldn't that match the Agilent results albeit within the limitations of its bandwidth?
-
Yesterday and today I did some further testing with the Rigol MSO1074Z-S only. This test uses the same 21:1 DIY probe and a specially made cable with the scope end internally terminated with an 0603 49.9 ohm resistor. This improved things, but I also noticed that reducing the sensitivity on the scope also improves the rise time significantly, I assume because it reduces the slew rate of the ADC driver opamp. It makes for a more reasonable, more "square" result.
Is there anything in the Rigol datacheet or operations manual that (a) indicates that occurs an/or (b) implies a reason?
I can't imagine a Tek or HP/Agilent/Keysight scope not mentioning such a phenomenon -
Since doing the test I am wondering if it's either to do with the relatively high transient, or reflections between the inline 50 ohm and the scope's high impedance front end. The coax added a much higher voltage to just using the 2pF cap on the Jim Williams base design.. When I get a moment I will investigate further. (By the way, I forgot to take the 10x off again on the Rigol, it's not really 340Vpp!.
-
...it's why we shouldn't be using BNCs!
Also, not all BNCs are created equal, and that too becomes quite apparent when your giga hertz.
I like SMB and SMC connectors for GHz applications. I've hacked up a few SMBs to make probes, and I'm sure RobRenz could make up some really nice ones out of 'em.
-
I can't imagine a Tek or HP/Agilent/Keysight scope not mentioning such a phenomenon
Indeed, my 475 for instance notes the rise/fall time is notably better for e.g. 2-3 div height than 6-8 div full screen. That's for an entire analog signal path, of course, but similar reasons could very well apply for the digital scope's preamp.
Digital scopes these days often use 0/5V or -5/+5V supplies for the analog bits, which means you'll inevitably see some slew rate limiting, compression or bandwidth limiting for large signals near the rails. Nice thing about the old stuff is the wide supply rails meant freedom from that (unless you seriously overdrive the input, in which case saturation recovery may be good or awful depending on make).
Tim -
I can't imagine a Tek or HP/Agilent/Keysight scope not mentioning such a phenomenon
Indeed, my 475 for instance notes the rise/fall time is notably better for e.g. 2-3 div height than 6-8 div full screen. That's for an entire analog signal path, of course, but similar reasons could very well apply for the digital scope's preamp.
Digital scopes these days often use 0/5V or -5/+5V supplies for the analog bits, which means you'll inevitably see some slew rate limiting, compression or bandwidth limiting for large signals near the rails. Nice thing about the old stuff is the wide supply rails meant freedom from that (unless you seriously overdrive the input, in which case saturation recovery may be good or awful depending on make).
Your analysis appears sound. My HP1740, which presumably predates your 475, only has to note that the 1mV/div bandwidth is a reduced 40MHz.
Back in the mid 70s HP delivered a minicomputer that didn't quite meet its published specifications. When Packard heard about it he sent a memo to the project manager asking him to ensure that in future the products did meet their specifications. The manager framed the memo, hung it on his wall, and the story and attitude became baked into HP's corporate folklore.
That was on reason why HP's products were trusted implicitly, in comparason with their competiton. (Tektronix excluded, of course, since they had similar attitudes.)
Not all change is for the better. -
I re-did the Jim Williams test. Firstly, I noticed today that the SMA-BNC with integral 50 ohm resistor wasn't 50 ohms. I don't know if it was physical or electrical stress (I was pumping quite a few volts into that 0603 49.9 ohm resistor albeit for only a few ns) but it was open anyway. I think we should forget that waveform as operator error on my part for now. Anyway, I fixed he internal 50 ohm termination and have re-run the tests, this time with an inline SMA 20dB pad at all times and with a reduced coax length on the 2pF cap too of about 50cm.
Rigol MSO1074Z-S, 500mV/div, 1X probe, 5ns/div
Rigol MSO1074Z-S, 200mV/div, 1X probe, 5ns/div. I was careful to ensure that the signal wasn't (apparently!) saturating. Note the very significant increase in rise time. You also get a relay click at the attenuator change form 500mV/div.
Agilent 54831D, 500mV/div, 1x probe, 10ns/div
Agilent 54831D, 500mV/div, 1x probe, 1ns/div
Agilent 54831D, 500mV/div, 1x probe, 1ns/div, direct connection to scope, internal 50 ohm load
Something else I've noticed on the Rigol is that even on a single shot stopped waveform, the measured Vpp value also changes as you change the vertical scale. I tried to reproduce the anomaly regarding the 200mV to 500mV rise time measurement above with an RF signal generator terminated to 50 ohm at the scope, and couldn't, so I'm not sure what's happening with the JW as a pulse source on the Rigol in that scenario.
-
I've noticed that you only have 5 sample points per division on the Rigol. So, it's trying to make a rise/fall time calculation with only a point or two in the transition area.
It could be doing the sin(x)/s reconstruction poorly with this few points. If possible, perhaps try turning off interpolation or switch to dots mode to see what you're actually getting. You could also download the raw data and examine it that way.
Or maybe it's just doing a bad job doing the rise/fall time calculation. If it has measurement cursors, try to see where it's anchoring its measurements. Or do the measurement manually if you can zoom the horizontal any further. You've chopped off the top of the 200mV/div waveform so I can't really tell if 3.5ns is right.
Are you doing this single shot? Do an average, or look at the rise/fall statistics (std deviation) to see if it's all over the place. It may also be dependent on where trigger point is occurring and where the samples land.
-
If possible, perhaps try turning off interpolation or switch to dots mode to see what you're actually getting. You could also download the raw data and examine it that way.
Always worth playing around with that, and it is usually my preferred display mode. It is easy for the eye/brain to "fill in the gaps".
I've seen cases where turning off interpolation revealed the waveform to be more repeatable than implied by the interpolation. -
You've chopped off the top of the 200mV/div waveform so I can't really tell if 3.5ns is right.
Maybe it's just under the OSD? -
Indeed, my 475 for instance notes the rise/fall time is notably better for e.g. 2-3 div height than 6-8 div full screen. That's for an entire analog signal path, of course, but similar reasons could very well apply for the digital scope's preamp.
Where is this noted? I looked through the 475 documentation and did not find it. I have seen this sort of specification before but it is insignificant on all of the oscilloscopes I have used. Almost all analog oscilloscopes would be using 5 divisions anyway for a rise/fall time measurement.QuoteDigital scopes these days often use 0/5V or -5/+5V supplies for the analog bits, which means you'll inevitably see some slew rate limiting, compression or bandwidth limiting for large signals near the rails. Nice thing about the old stuff is the wide supply rails meant freedom from that (unless you seriously overdrive the input, in which case saturation recovery may be good or awful depending on make).
0/+5.0 to 0/+1.8 volts is typical now depending on the stage for the analog supply but the signal path is limited to below 2 volts differential peak-to-peak. I do not know what preamplifier the Rigol 1000z series uses but the ADC is a Hittite HMCAD1511 (1.8 volts) which supports a 2 volt peak-to-peak differential input. The 2000A series uses the National/TI LMH6518 amplifier (5 volts) and ADC08D500 ADC (1.8 volts) which supports a peak-to-peak input below 1 volt.
None of these parts specify a slew rate limit as such although it could be estimated from their full power bandwidth but that yields numbers in excess of what is required. The ADCs naturally do not specify overdrive recovery but the LMH6518 preamplifier has a specified faster overdrive recovery than its settling time which is pretty neat. Overdrive recovery would almost certainly be limited by the input buffer but some DSOs have horrendous overdrive recovery in later stages.
-
Indeed, my 475 for instance notes the rise/fall time is notably better for e.g. 2-3 div height than 6-8 div full screen. That's for an entire analog signal path, of course, but similar reasons could very well apply for the digital scope's preamp.
Where is this noted? I looked through the 475 documentation and did not find it. I have seen this sort of specification before but it is insignificant on all of the oscilloscopes I have used. Almost all analog oscilloscopes would be using 5 divisions anyway for a rise/fall time measurement.
Hmm, looking through my documents, I don't see it either...
Maybe I conflated that with something I read elsewhere, perhaps a lesser quality scope.
Tim
