Attenuation: | X1 : X10 |
Input Resistance: | 1Mohm ±2%(X1) : 10Mohm±2%(X10) |
Input Capacitance: | 1Mohm ±2%(X1) : 10Mohm±2%(X10) |
System bandwidth: | DC~100MHz |
Max input Voltage: | X1: 300VDC+Peak AC X10: 600VDC+Peak AC |
Net Weight: | 55g |
Cable Length: | 120cm / 42.2 inches |
Dave.S you would think it’s a huge oscilloscope specially on that picture, but it is one of the compact oscilloscopes in its time, you remember my newly bought Tek 2221A right?So it's compact longitudinal-ly ? ;D
This Iwatsu is only 6kg (13,2 pounds) and easier to carry around and Tek 2221A is 8.2kg (18 pounds) without its pouch on top of it ;)
I like my Iwatsu, the Volts/Div and Time/Div, buttons need some maintenance time to time and after that it works perfect again.
What's the loading at 200/300 MHz for these supposedly 200/300 MHz probes? How do they perform when driven frome a higher source impedance?
The specification says the input capacitance is 18.5pF to 22.5pF, which is quite similar to Tektronix and Agilent probes.300-500 MHz bandwidth Agilent/Tek probes would have something like 8-12 pF of input impedance, or about half of the HF loading.
I set up a network analyzer to sweep 100 kHz to 300 MHz and display impedance on a Smith chart.How accurate would a network analyzer be for this measurement, given that the impedance is very far from 50 ohm?
You notice that the rise time of my 300 MHz scope was 1 nS when the tunnel diode pulser was connected directly to the scope. The rise time through the Chinese probes was also essentially 1 nS, so I would need a faster scope to truly measure the probe rise time, which must be better than 1 nS.Rise time doesn't tell the whole story, however. The cheap probes showed substantial overshoot, much more than with the direct connection or Agilent probe, indicating an over-peaked response. This results in a higher apparent bandwidth, at the cost of transient response (accuracy at and shortly after the rising edge). A CW sweep of the probe across the scope's bandwidth might show a dip in the frequency response somewhere far below 300 MHz.
For a price of 10 to 15 dollars, these probes are amazingly good. The build quality is good and I keep the Agilent 10073C probes put away and just use the Chinese probes for everyday use.I can see why you'd rather kill a $10 probe than a $200 probe, but I would still keep some good probes around for when accuracy matters or you don't trust the cheapies for some reason. My experiences with durability of the cheap probes and their accessories (eg. Rigol) aren't too good, but I guess for $10-15 you don't care.
The specification says the input capacitance is 18.5pF to 22.5pF, which is quite similar to Tektronix and Agilent probes.300-500 MHz bandwidth Agilent/Tek probes would have something like 8-12 pF of input impedance, or about half of the HF loading.
I measured the 500 MHz Agilent 10073C at low frequencies and its input capacitance is 12.55pF compared to their spec of 12 pF. The Chinese probe measured about 16 pF, better than their spec. However, at high frequencies, transmission line effects begin to dominate. At 300 MHz the impedance of the 10073C 500 MHz probe is 42+j74 which works out to about 5 pF in parallel with 176 ohms.
I have destroyed one probe over the course of 2 years, and I made some measurements on the cable. The center conductor is very fine resistance wire, with a DC resistance of about 230 ohms end to end.I set up a network analyzer to sweep 100 kHz to 300 MHz and display impedance on a Smith chart.How accurate would a network analyzer be for this measurement, given that the impedance is very far from 50 ohm?
Perhaps you are speaking of the 100 kHz impedance of .01 + j7667 ohms, on the far right edge of the Smith chart. In the 100s of megahertz range, the trace is in the region of the Smith chart where the impedances are on the order of 100s of ohms and the accuracy is quite good.You notice that the rise time of my 300 MHz scope was 1 nS when the tunnel diode pulser was connected directly to the scope. The rise time through the Chinese probes was also essentially 1 nS, so I would need a faster scope to truly measure the probe rise time, which must be better than 1 nS.Rise time doesn't tell the whole story, however. The cheap probes showed substantial overshoot, much more than with the direct connection or Agilent probe, indicating an over-peaked response. This results in a higher apparent bandwidth, at the cost of transient response (accuracy at and shortly after the rising edge). A CW sweep of the probe across the scope's bandwidth might show a dip in the frequency response somewhere far below 300 MHz.
The direct connection to the scope through a 50 ohm feedthrough and the scope set for 1 M ohm input impedance showed some overshoot; I don't know if the scope exhibits some overshoot or if the tunnel diode pulser has some. The P6100 probe showed somewhat less overshoot than that. The Agilent 10073C probe showed almost no overshoot but the direct connection to the scope did show overshoot; what do we make of that? At rise times in the nanosecond range, rather than 125 pS, the Chinese probes don't exhibit overshoot.
I performed sweeps a while ago; I'll post those.
I would expect the performance with a fast high-amplitude signal to be substantially worse than the Agilent probe, since capacitors and resistors in the attenuation network will likely have some voltage coefficient.
The capacitors are low pF caps, and there would be no need to use high-K dielectrics which exhibit a large voltage coefficient. The caps are most likely NPO which has a negligible voltage coefficient. I would also expect the resistors to have negligible voltage coefficient. I don't think that "...some voltage coefficient" necessarily translates into "...performance with a fast high-amplitude signal to be substantially worse..."
At any rate, the largest very fast rise pulse I have available to me is a 900 pS 5 volt pulse from a 74AC series logic buffer, and I didn't see any worsening of performance. For a somewhat slower rise, say 3 or 4 nanoseconds, I can generate a 100 volt pulse, and I didn't see any worsening of performance there either.For a price of 10 to 15 dollars, these probes are amazingly good. The build quality is good and I keep the Agilent 10073C probes put away and just use the Chinese probes for everyday use.I can see why you'd rather kill a $10 probe than a $200 probe, but I would still keep some good probes around for when accuracy matters or you don't trust the cheapies for some reason. My experiences with durability of the cheap probes and their accessories (eg. Rigol) aren't too good, but I guess for $10-15 you don't care.
I broke one of the Chinese probes after about 2 years of use by accidentally yanking it, which is the same thing that has killed the expensive probes in the past. I'm quite satisfied with the Chinese probes.
The direct connection to the scope through a 50 ohm feedthrough and the scope set for 1 M ohm input impedance showed some overshoot; I don't know if the scope exhibits some overshoot or if the tunnel diode pulser has some. The P6100 probe showed somewhat less overshoot than that. The Agilent 10073C probe showed almost no overshoot but the direct connection to the scope did show overshoot; what do we make of that? At rise times in the nanosecond range, rather than 125 pS, the Chinese probes don't exhibit overshoot.I would tend to believe the direct 50 ohm connection, since 50 ohm inputs tend to be superior to high-Z inputs. I would also expect the pulser to be happier since the load is much more resistive. On the other hand, it's quite common for probes to compensate for imperfections in the scope's input impedance and response. This is why companies used to make several different probes with the same bandwidth. Nowadays they tend to use the same front-end for most of their scopes, reducing the number of different probes they need to make.
The capacitors are low pF caps, and there would be no need to use high-K dielectrics which exhibit a large voltage coefficient. The caps are most likely NPO which has a negligible voltage coefficient. I would also expect the resistors to have negligible voltage coefficient. I don't think that "...some voltage coefficient" necessarily translates into "...performance with a fast high-amplitude signal to be substantially worse..."The big scope vendors like Agilent (HP) and Tektronix used to spend a lot of time and effort to reduce voltage and frequency-dependent effects in the attenuation hybrid. I'm not sure if this technology is within reach for an Asian manufacturer that makes $15 probes, maybe it is.
Attached are images of the swept frequency response of the DSO5034 scope, the Chinese probes and the Agilent 10073C 500 MHz probe. The probe sweeps were displayed on the DSO5034 scope, so take the rolloff of the scope into account when considering the probe sweeps.Thanks for posting these. Do you expect the setup to cause any significant errors? Eg. reflections from the not very well terminated transmission line?
Just for grins I connected the cable from the damaged Chinese probe to the network analyzer and shorted the end. The result is shown in the attached image. The marker is at 100 kHz and you can see that the impedance at that frequency is about 230 ohms real with a small capacitive part.Neat! Not sure what to conclude from this, however, pretty much what you would expect from a lossy transmission line I guess.
The direct connection to the scope through a 50 ohm feedthrough and the scope set for 1 M ohm input impedance showed some overshoot; I don't know if the scope exhibits some overshoot or if the tunnel diode pulser has some. The P6100 probe showed somewhat less overshoot than that. The Agilent 10073C probe showed almost no overshoot but the direct connection to the scope did show overshoot; what do we make of that? At rise times in the nanosecond range, rather than 125 pS, the Chinese probes don't exhibit overshoot.I would tend to believe the direct 50 ohm connection, since 50 ohm inputs tend to be superior to high-Z inputs. I would also expect the pulser to be happier since the load is much more resistive. On the other hand, it's quite common for probes to compensate for imperfections in the scope's input impedance and response. This is why companies used to make several different probes with the same bandwidth. Nowadays they tend to use the same front-end for most of their scopes, reducing the number of different probes they need to make.
The pulser saw 50 ohms in both cases. When the scope was set for 1M input, the pulser was connected to the scope through a 50 ohm feedthrough, the sort you use to terminate the older Tektronix current probes.
The pulser is in a small metal box, and no cable was used to connect it directly to the scope; this is the pulser: http://www.ebay.com/itm/Tektronix-067-0681-01-Tunnel-Diode-Pulser-calibration-fixture-/110847494323?pt=LH_DefaultDomain_0&hash=item19cf0688b3 (http://www.ebay.com/itm/Tektronix-067-0681-01-Tunnel-Diode-Pulser-calibration-fixture-/110847494323?pt=LH_DefaultDomain_0&hash=item19cf0688b3)The capacitors are low pF caps, and there would be no need to use high-K dielectrics which exhibit a large voltage coefficient. The caps are most likely NPO which has a negligible voltage coefficient. I would also expect the resistors to have negligible voltage coefficient. I don't think that "...some voltage coefficient" necessarily translates into "...performance with a fast high-amplitude signal to be substantially worse..."The big scope vendors like Agilent (HP) and Tektronix used to spend a lot of time and effort to reduce voltage and frequency-dependent effects in the attenuation hybrid. I'm not sure if this technology is within reach for an Asian manufacturer that makes $15 probes, maybe it is.
I think reducing the voltage dependencies due to voltage coefficient in a small NPO trimmer can't be too hard. Frequency dependencies are another matter. Since Tektronix did the work a long time ago, all the Chinese manufacturer has to is copy.
Anyway, the measurements I've been able to make show that for rise times I care about (not 125 pS, 100 volt excursions), a few nanoseconds, I don't see any noticeable performance degradations that I would attribute to voltage coefficient problems. When I compensate the probe for a pulse amplitude of 1 volt, and then apply a 100 volt pulse, I don't see any changes in the displayed pulse edge shape.Attached are images of the swept frequency response of the DSO5034 scope, the Chinese probes and the Agilent 10073C 500 MHz probe. The probe sweeps were displayed on the DSO5034 scope, so take the rolloff of the scope into account when considering the probe sweeps.Thanks for posting these. Do you expect the setup to cause any significant errors? Eg. reflections from the not very well terminated transmission line?
The probes were connected to the scope with the scope input set to 1M ohm. The probe tips were poked into the center conductor of the type N connector which was the sweep generator output, with a piece of braid wrapped around the probe tip ground collar and the shell of the type N connector to provide reasonable continuity there. Whatever errors there may have been, due to reflections in the probe cable were what one would have to tolerate if one were probing such high frequencies without maintaining a 50 ohm environment, but rather using a 10x probe.
It looks like the DSO5034 + 10073C does not meet specs, the -3 dB point appears to be about 250 MHz to me. Is this combination recommended by Agilent? The PP-150 response looks quite horrible to me, with the response up +3dB from 100 MHz to 200 MHz. The P6100 looks quite good, however. This is consistent with the step response measurements.
The 10073C is the probe provided with the DSO5034 and also the 500 MHz DSO5054. I was puzzled by the less than expected frequency response. The probe is spec'd to have a 700 pS rise time, but I can't verify that. It certainly does well at 1 nS, and has a nice gentle rolloff in frequency response.
I'm surprised that these probes exceed the specs by so much, especially switchable (1x/10x) probes. For the price, I wouldn't even expect them to meet their specs. Would the quality control be insufficient to meet this performance consistently, or would there be some other issue that we're missing?
I can't explain it, but when I discovered their performance, I was also surprised. Even if they only last a year, at that price they're great for everyday work, allowing one to preserve the expensive probes that came with the scope for special occasions.
When searching eBay for P6100 I find the probe under various brands: Yang Xun, no brand at all, Caltek. Any idea if they're all made by the same manufacturer and relabeled or if P6100 is the generic name for scope probe in China?
I can't answer that, but given the ability of the forum readership to discover things like that, I expect someone to let us know.Just for grins I connected the cable from the damaged Chinese probe to the network analyzer and shorted the end. The result is shown in the attached image. The marker is at 100 kHz and you can see that the impedance at that frequency is about 230 ohms real with a small capacitive part.Neat! Not sure what to conclude from this, however, pretty much what you would expect from a lossy transmission line I guess.
The thing the network analyzer curves show is that the input capacitance specified by the manufacturer is a (relatively) low frequency spec. At high frequencies, the input capacitance is determined by the cable; the compensation trimmer and scope input capacitance are decoupled from the probe tip by the properties of the cable, which is several wavelengths long at 100s of megahertz.
The pulser saw 50 ohms in both cases. When the scope was set for 1M input, the pulser was connected to the scope through a 50 ohm feedthrough, the sort you use to terminate the older Tektronix current probes.In one case it saw a resistive 50 ohm, in the other case 50 ohm // 10-20 pF. The capacitance becomes quite significant at the frequencies generated by a 125 ps rising edge.
The probes were connected to the scope with the scope input set to 1M ohm. The probe tips were poked into the center conductor of the type N connector which was the sweep generator output, with a piece of braid wrapped around the probe tip ground collar and the shell of the type N connector to provide reasonable continuity there. Whatever errors there may have been, due to reflections in the probe cable were what one would have to tolerate if one were probing such high frequencies without maintaining a 50 ohm environment, but rather using a 10x probe.I was thinking of standing waves in the cable to the probe (not much of a cable it seems), or the leveling circuit of the generator being influenced by the reactive load/reflections it sees. I know this is a problem in (fast) scope bandwidth testing, and the reason why they use as many attenuators as possible between the generator and the scope. Even those '50 ohm' input scopes often don't have great VSWR specs.
Rohde & Schwarz have a new scope out, and I've been seeing advertisements touting their Gaussian rolloff, giving good pulse fidelity.Funny. Agilent published an appnote about the superiority of the brick wall response for pulse measurements a number of years ago ;). I believe their argument was that the superior pulse fidelity of the Gaussian roll-off is only relevant if the pulse is faster than the scope, and you should just buy a faster scope if that's the case.
The pulser saw 50 ohms in both cases. When the scope was set for 1M input, the pulser was connected to the scope through a 50 ohm feedthrough, the sort you use to terminate the older Tektronix current probes.In one case it saw a resistive 50 ohm, in the other case 50 ohm // 10-20 pF. The capacitance becomes quite significant at the frequencies generated by a 125 ps rising edge.
In your very next comment, you say "Even those '50 ohm' input scopes often don't have great VSWR specs.", so we don't know to what degree "...it saw a resistive 50 ohm". I won't be back home for a few days, but when I get back I will connect the 50 ohm input of the scope to the network analyzer and see just how good a 50 ohm load it is. For all we know even the 50 ohm input may look significantly capacitive to a 125 pS edge.The probes were connected to the scope with the scope input set to 1M ohm. The probe tips were poked into the center conductor of the type N connector which was the sweep generator output, with a piece of braid wrapped around the probe tip ground collar and the shell of the type N connector to provide reasonable continuity there. Whatever errors there may have been, due to reflections in the probe cable were what one would have to tolerate if one were probing such high frequencies without maintaining a 50 ohm environment, but rather using a 10x probe.I was thinking of standing waves in the cable to the probe (not much of a cable it seems), or the leveling circuit of the generator being influenced by the reactive load/reflections it sees. I know this is a problem in (fast) scope bandwidth testing, and the reason why they use as many attenuators as possible between the generator and the scope. Even those '50 ohm' input scopes often don't have great VSWR specs.
Knowing that a passive 10x scope probe is not going to be well matched to a 50 ohm environment, I wanted to do the best I could but in a way that would be typical of how an ordinary user might (mis)use the probe, so I connected the probe tip directly to the generator output connector with no intervening cable. I don't see the sort of undesirable behaviors I would expect to see if there were substantial reflections, or a leveling circuit failing to do its job. The scope cable is so lossy that any reflections will be greatly attenuated.
At any rate, since a 10x scope probe is not a controlled impedance cable, how else can we measure its "frequency response"?Rohde & Schwarz have a new scope out, and I've been seeing advertisements touting their Gaussian rolloff, giving good pulse fidelity.Funny. Agilent published an appnote about the superiority of the brick wall response for pulse measurements a number of years ago ;). I believe their argument was that the superior pulse fidelity of the Gaussian roll-off is only relevant if the pulse is faster than the scope, and you should just buy a faster scope if that's the case.
Superior in what way?
In your very next comment, you say "Even those '50 ohm' input scopes often don't have great VSWR specs.", so we don't know to what degree "...it saw a resistive 50 ohm". I won't be back home for a few days, but when I get back I will connect the 50 ohm input of the scope to the network analyzer and see just how good a 50 ohm load it is. For all we know even the 50 ohm input may look significantly capacitive to a 125 pS edge.Neither is perfect, but I expect the scope input to be significantly better, unless the designers were really lazy and just switched a 50 ohm resistor across the input. Hard data trumps any assumptions, however.
I don't see the sort of undesirable behaviors I would expect to see if there were substantial reflections, or a leveling circuit failing to do its job. The scope cable is so lossy that any reflections will be greatly attenuated.Fair enough, I'm just trying to exclude any potential source of uncertainty. The usual method would be to terminate the 50 ohm connection as close to the probe tip as possible, to minimize the sub length, but you're probably already quite close to this.
At any rate, since a 10x scope probe is not a controlled impedance cable, how else can we measure its "frequency response"?
Superior in what way?Rise time measurements. This (http://cp.literature.agilent.com/litweb/pdf/5988-8008EN.pdf) is the appnote I was referring to.