Spectrum Analysis Comparing Gen II vs III

Do you have a sweep without the vehicle (aka., Farady cage?) for the background?

Vehicle in "Ready" and "Park"?

Any change if in "Ready", "D" with parking brake set or wheels chocked?

Thanks,
Bob Wilson

 

I would have asked how you got the Priuses into a screen room, but the FM signals answer that.

 

Not one that will hold a car.

 

Snijd,

Aha! And thank you!

Wonderful - you've got a 2006 and a 2010 Prius, in roughly the same environment.

I have a bunch of comments. If possible, as a results of these comments, we (as a group) might ask for a retest with some slightly different settings; your call if it's difficult to get both cars where you can lay your hands on them.

First: I note that with both cars off, the noise floor of the system, when centered at 7.5 MHz, is around -90 dBm. The noise floor, with the whip attached, at 100 MHz centered is around -80 dBm. Hmm.. Can you tell us what the noise floor of the spectrum analyzer is with a 50 ohm termination attached directly to the input of the spectrum analyzer, set up the way you had it, at both frequencies? I'm interested in the analyzer's noise floor.

Second: I notice that in both of the settings you have the spectrum analyzer going from "zero" frequency to twice the frequency at the center. Umm... I understand your wish to cover as much ground as possible, but I wonder at the resolution of the spectrum analyzer. Yes, you've got the video bandwidth set to 30 kHz, which is nice and narrow for the application intended, but I don't think I would have set it up quite that wide, myself.

For one thing, the screen on this thing looks decidedly digital. I may be very wrong, and my apologies if I am. But, if it is digital, there's an implication that with the RF span set as wide as you have it, spurs that don't "line up" with the resolution of the screen may not show. Part of the reason I'm suspicious about this is that there are some whopper signal generators in the MF/HF band - and I don't exactly see them on any of your scans. One of the other posters had a valid comment about ye olde-tyme Faraday cage, and I understand you're not having one in your back pocket , but it's the lack of other radiators that bother me.

It looks like, based upon your readings, that there's a motor switching frequency around 7.5 MHz. Nice. Hmm.. That's near the top of the amateur radio 7-7.3 MHz. My memory says there used to be a lot of long-range HF broadcast stations around those parts. Bet somebody got irritated with Priuses interfering with the BBC and the like.

Given that we've all heard a lot of kvetching about the FM radio band, I would love it if you would center the spectrum analyzer on 100 MHz with a span of 4 MHz, then take one reading with the car on and another with the car off. FM stations have a span of 200 kHz; with a bit of luck, one should be able to see the "clear" spots where there aren't any local stations and the strong spots where there are, and turning the car on and off should show up any EMI in and around the range.

Comments?

KBeck.

 

Shouldn't this be a function of the resolution bandwidth? The former are taken at 30kHz, the later at 300kHz, which should account for a 10 dB difference is displayed noise floor.

 

Snijd,

Now we're cooking with gas. First, the one "error" you've got in there. 88 to 108 MHz with the resolution bandwidth set to 300 kHz probably wasn't that great an idea, sorry. 30 kHz would have given a much better view, in that you would have been able to make out the individual radio stations.

But, now the good stuff. Those two pictures centered on 98 MHz with 30 kHz bandwidth, with an "off" car and an "on" car show a heck of a lot of noise riding on top of the RF signal. Unless I'm missing something, it looks like the noise signal is within 10 dB of the radio! That, in my humble opinion, is 'way too much.

Over in the Audio and Electronics forum, Jeremy successfully measured quite a bit of RF interference coming out of a turned-on Prius. In addition, he found a strange one: When he turned on the windshield wipers (!) the interfering signal dropped significantly when the wiper arms were pointing straight up.

OK. In previous posts I was doubting that there was that much interference coming out of the Prius. It may still be there or not on a car-by-car basis. I have got to get to the FCC rule 97 stuff on emissions. Later today.

KBeck

 

Do remember that you cannot directly compare noise floors at different center frequencies unless the bandwidths are set the same. If the bandwidth is hardwired to the span, then you need to set the spans the same.

On the pictured displays, I'm not seeing any indication of averaging multiple measurements, and the 40 db spread on the individual noise bins suggests its absence. That leaves a lot of visual hash, with a mental peak-detecting effect on the display raising the apparent noise floor. You can nail down it more precisely, and find a lot more weak signal detail, if averaging is turned on.

 

Most spectrum analyzers allow one to set the span, the resolution bandwidth, and the scan rate. For a given span, the smaller the resolution bandwidth, the slower the scan rate.

In most of the spectrum analyzers I've used, the resolution bandwidth defaults to some number when the span is set. If you want a narrower resolution bandwidth, one can normally do so, but the scan rate slows down.

I agree than a 2 kHz bandwidth over, say, a 400 kHz span centered on an FM station would be very useful, with the car both on and off. That should let us see both the clear area outside the station (when the car is off) and how much actual interference is being generated (when the car is on).

Using a narrower bandwidth has the additional helpful characteristic of limiting the noise power. Narrower the bandwidth, less noise (especially that of the spectrum analyzer) gets through.

KBeck.

 

Sorry, I once had occasion to deal with a Faraday cage in 'real life'. Metal buildings are not that uncommon but apparently not available in this case. Just one random thought about the images.

2006 inverter, 7.5MHz center frequency. Sensing antenna was situated next to the inverter, under the hood.

The black, areas under the curve share the characteristic of harmonic spread. In particular, I would want to read out the black peaks frequency spikes under the curve and see if we can back into the primary frequency or frequencies.

I started to post about instrumentation methodology when this random thought occurred:

What are we after in these frequency ranges?
​

When I first got my NHW11, I fabricated sensor coils and used AudioCity to record power to and from MG1 and MG2. Had I continued, I would have gone to Hall effect sensors to read out current and frequencies along with applied voltages from the coils (I still have the parts.) This is because I'm interested in Prius engineering data and understanding power flows. For example, the fastest MG1 speed is reported ~10,000 rpm for a three phase, AC motor, we're looking at 30 kHz, square waves. I suspect the ZVW30 has even higher MG1 rpm and primary power frequencies when in operation.

The reason I bring this up is with this spectrum analyzer running up in the megahertz where at best we may see very high frequency harmonics from the square wave, power circuits but why do we care? Understand this is not a hostile or challenge question but I'm trying to understand the use of anything beyond the 2nd or 3d harmonic?

Bob Wilson

 

OK somebody with, and seems to know how to use a Analyser. Pic 1 does show lots of Bradband noise, but at -79 dB, thats not bad. Pics 2 and 3 are exacty what I would expect to see in the middle of the FM Broadcast band, so OK there. The noise at 7.5 MHz is kinda irrelevent, uness you want to work 40 meter SSB or CW in the car. Picture 5 is a REALLY important shot Notice how the FM carriers are all well defined and symetrical.. This is a proper set of signals for a radio to make use of. Pic 6 sae thing. If these pics was taken off the antenna cable that feeds the Prius radio, this antenna system is working PERFECTLY, and would sound GREAT!!! Lets jump ahead to pics 12 and 13. Here I really wish you would have included a pic from the 2010 antenna with a sweep span of 20 Mhz like you had in pic 5. However in comparison the big difference is the Gen II has 5 or so more dBm of signal to stick into the radio with -54.36 dB level of signal, wheeas the 2010 has on a -49,11 dBm. I woud really like to see the 2010 Antenna out with a 20 meg sweep or less pic to see how clean the carriers are and how well the antenna and preamp are handling the RF!
Nice job on the pics. They really are quite useful!!

 

Not at all! Gaining insights is important and the impact on HF should be significant. Heck, I'm surprised the AM band doesn't 'go nuts' although considering the quality of programing . . . it might be an improvement.

Knowing the goal is ham radio operation simplifies a lot and suggests methodologies. For example, mapping the MG1 and MG2 primary frequencies and then using the harmonics to identify the ones most likely to impact operation the relevant bands . . . assuming we're looking at mobil operation. For fixed operation, the current approach works.

I concur on the great photos. I wasn't trying to be skeptical as much as suggesting we might improve the methodology if specific goals and objectives are identified. My personal interest is in power systems and flow and not so much in RF noise and bands. There is a relationship but somewhat loose.

Bob Wilson

 

Excellent analysis

 

There is a big difference between a Faraday shield room and an anechoic chamber. Are you sure you all aren't confusing the difference between the two?

Can't see why an average Prius owner would care about shield room data. Shield rooms can be noisy and uncomfortable, anechoic chambers can be creepy because most people haven't experienced such quiet.

Big difference between acoustic noise and higher frequency noise.