rrTV-PHOTO   New HD TV
HOME   rrTV-PHOTO   GALLERIES   MY GALLERY   HELP-FAQ
myHOME PM pmRR MEMBERS 343 ONLINE 19 EVENTS SEARCH REGISTER  START HERE
 
5 pages [ <<    <    ( 1 )     2      3     NEXT    >> ]8353 viewsPOST REPLY
Fast Lad Performance . Ace Hobby . Esprit Model

.
.
Main Discussion > airtronics "worlds fastest"??
 
 
w.pasman
Elite Veteran
Location: Netherlands

I saw the airtronics transmitters to be often quoted here for their "world's fastest PCM 1024 response time". However I found nothing substantiating that claim. Does someone have technical specifications of their modulation scheme, latency measurements or something else?
10-02-2003 Over year old.
HOMEPAGE  
 
 
vetrider
Elite Veteran
Location: Daleville, AL (Ft.Rucker)

I believe it has to do with the sampling rate of the signal but don't hold me to it.

It was mentioned in a post a week or two ago.

Nolan Manley
RREMODELS, KBDDintl, Magnum Fuel
Stratus, ODIN, Knight3D, ATOM 500, TREX 700N.
10-02-2003 Over year old.
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Hi, W.Pasman!

I dug into this a bit, and in the process, discovered that Airtronics no longer has the world's fastest PCM - Hitec does!

In the newly revised Hitec Eclipse 7, Hitec claims a "response time" of 11 milliseconds, which they say is "nearly 50% faster than any other PCM technology". If that's the case, then the response time of Airtronics is probably on the order of 20 milliseconds. Unfortunately, I couldn't find comparable figures for JR and Futaba, but they're no doubt even higher.

The way I translate this (which may or may not be correct) is that the "response time" is really the main loop time - that is, the amount of time it takes the transmitter to completely cycle through each of its inputs, compute the relevant mixing, and transmit a complete PCM frame to the receiver. As VetRider suggests, this sets the effective sampling rate for each input, or about 90 Hz for Hitec, 50 Hz for Airtronics, and less for current JR and Futaba systems.

Assuming that this is correct, then response time is a function of the CPU speed in the transmitter, and the efficiency of the code that's running on it. We do know that that's an issue, especially on high-end radios with complex mixing capabilities. If you remember the update to the 9Z WC2, Futaba revised their software and claimed that they were able to reduce the response time near center stick (due to the simplicity of byte math, as compared to maintaining 10-bit precision on an 8-bit processor).

The good news is that Moore's Law applies to the CPUs in transmitters as much as it does to any other CPU, so we should continue to see response times fall.

While it wasn't quite what you were looking for, I hope this helps!

Cheers!
MarkF
10-02-2003 Over year old.
HOMEPAGE  
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Howdy!

While those sampling rate differences may seem like a very big deal, it's important to note that nothing happens until the servo reaches its target rotation. Given that we're talking about speed differences of 0.01-0.02 seconds between these transmitters, for most of us, servo speeds overwhelmingly determine the true response time.

Interestingly enough, Airtronics does sell the world's fastest commercial R/C servo, the 94758Z, at 0.06 seconds. If you're a competition flyer, and you're running super servos, then the transmitter response time begins to become an issue.... assuming, of course, that your brain and your thumbs are fast enough to take advantage of it!

Cheers!
MarkF
10-02-2003 Over year old.
HOMEPAGE  
 
 
Optech
Key Veteran
Location: Vista/Oceanside, CA.

Actually, Airtronics claims a 200 microsecond response time which would be 0.2 milliseconds. Significantly faster than Hitecs 11milliseconds.

I suspect though that its an apples and oranges situation as nobody really knows the conditions or parameters of their testing.

Mike

Viva La Airtronics!
10-02-2003 Over year old.
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Hi, Optech!

That's definitely a goofy claim on Airtronics part. If that represented the loop time, then Airtronics would be updating their servos 5,000 times/second, which they're definitely not! Consider that the maximum theoretical number of bits/second on a 10 KHz channel with FSK keying is about 10K bits/second. Let's imagine that a 9 channel transmitter adds two control channels/frame (one for framing and type info, and one for parity), and that all 11 of these channels are 10 bits long, as required for PCM-1024. This would mean that a single frame would be 11 * 10 = 110 bits long.

Let's assume that the actual data rate is 10K bits/second. That means that it would take 11 milliseconds to send a single frame. Heh, heh - that's exactly the response time that Hitec claims!

So, I'll now modify my earlier statement: without more complex coding or modulation schemes, it'll be hard to significantly decrease response times, since we're now running up against a bandwidth limit, rather than a processing speed limit! It sounds like what Hitec has done is to make their processor run fast enough that it can compute a full frame while the previous frame is being transmitted.

Fortunately, the easiest way to gain a bandwidth savings would be to use a differential encoding scheme such as ADPCM - that'd buy at least a factor of two (and probably three).

I can't imagine what the heck Airtronics is claiming with that 200 uS number, since that's enough time to send exactly two bits from the transmitter to the receiver! Talk about marketing B.S.!

Cheers!
MarkF

P.S. Realistically, 8K bps is a more likely maximum throughput in a 10 KHz channel, depending on the signal/noise ratio and real world filters (this would correspond to an "alpha", or excess bandwidth figure of 0.25). However, Hitec can still achieve their result, since the Eclipse is a 7-channel transmitter, and most likely, the control channel won't have to be a full 10-bit field.
10-02-2003 Over year old.
HOMEPAGE  
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Hi, Optech!

On further thought, I'll bet that Airtronics is probably sending at a data rate of 5K bps. That means that their claimed "response time" is the time to send one bit from the transmitter to the receiver.

I hate marketing guys...

Have Fun!
MarkF
10-02-2003 Over year old.
HOMEPAGE  
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Hi!

David: I'm sorry that I don't have a Hitec, so I can't help you on that one. It might well be that Hitec's running real PCM, but they decided to economize on bandwidth by having the receiver store the failsafe settings. There's no technical advantage to sending failsafe values from the transmitter, other than ease of setup. In fact, you could argue that storing them in the receiver is actually better, since it eliminates the possibility of storing incorrect values in a high bit-error rate environment!

On another topic entirely, I thought I'd add a quick note on why fast response times are relevant. After all, human response time is on the order of 100 mS (i.e., a 100 mS response seems essentially "instantaneous"). However, when we can anticipate an action, such as when we are planning a maneuver, we are capable of much faster timing accuracies. For example, almost 25 years ago when I was writing keyboard processor code, I used to be able to stop a downcounter within +/-1 ms, and I'm not at all dexterous. So, put one of these fast systems in Alan Szabo Jr.'s hands, and he'll show you why fast response time is important! Is it any surprise that he runs with 94758Zs?

Cheers!
MarkF
10-02-2003 Over year old.
HOMEPAGE  
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Hi, David!

One more quick thought. If you're looking at the Hitec Eclipse manual on the web, it's probably just out of date. The first release of the Eclipse didn't support PCM, and they just may not have gotten around to updating the website with the new manual.

Good Luck!
MarkF
10-02-2003 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: City of California in the state of Maryland

Something else to consider is that the control signals from the receiver to the servos are only updated at 50 Hz. That is, the servos are told where to be/go every 20 ms. Only after receiving that signal can the servo begin to react to it.

- John
10-02-2003 Over year old.
 
 
tonysrepair
Veteran
Location: portland oregon

The hitec settings for qpcm is in the setup menu .( holding the two farthest left buttons down while turning the radio on) I won one of their qpcm radio's two month ago . The only problem is the reciever just failed , luckly on the ground . Not impressed with the quality. Compared to my 9C there is no contest . The 9C imput is much more precise. In my opinion
10-02-2003 Over year old.
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Hi, John!

Absolutely! While we're debunking, let's keep going! Any computer radio that does mixing must read all of the A/D converters before it can process a single channel, so it isn't like the transmitter can read the throttle control position just before sending it out. Instead, the transmitter has to build a complete frame at a time, and then transmit it. This means that the worst case latency out of the transmitter is more like two frame times, so it's even more relevant - with the Airtronics servos, that'd add up to 25% of the time-till-receiver!

Now, the $64,000 question is whether Hitec is claiming the frame rate is 11 mS, or the actual latency is 11 mS (not counting the servo frame latency)? If it's the latter, then they've already adopted a differential coding scheme, and their frame rate is more like 5mS! One of these days, I'll have to borrow an Eclipse and throw it on a digital scope...

Have Fun!
MarkF
10-02-2003 Over year old.
HOMEPAGE  
 
 
w.pasman
Elite Veteran
Location: Netherlands

Hi guys

Thanks for all the responses so far.

A few general remarks, to solve some misunderstandings here
1) I think latency should be measured between the moment you move the stick or flip the switch and the moment the receiver gives the signal corresponding to the new stick or switch. Thus, the overall latency. You can not include the servo in this, unless you say which servo you used and how much latency that part took.
2) Latency is related to the update rate (measurements per second) in the sense that you never can get lower latency than 1/the number of measurements per second. So if you do 50 stick measmts per second you will have a (worst case) latency of AT LEAST 20ms. However, that is EXCLUDING encoding, buffering, transmission, decoding, error correction etc. Same story holds for the number of frames that are sent per second. In the worst case, which is exactly what latency is about, all those cyclic things add up, unless the steps are synchronised.
3) Futaba's PCM 1024has 1 frame every 14,25ms, so that is at least their latency. Graupner/JR SPCMhas a frame of 22ms. This info is from http://www.aerodesign.de/peter/2000...CM_PPM_eng.html

4) can everyone quoting figures MENTION WHERE YOUR QUOTE COMES FROM? Then I can have a look there if there's more info to it and also estimate the credibility of the source.


MarkF
11ms that is pretty low "response time". Wonder what response time means, overall latency, frame time or something else? Where did you get this value from?

> While it wasn't quite what you were looking
> for, I hope this helps!
Sure, your addition to this makes this thread even more interesting!

> servo speeds overwhelmingly determine
> the true response time
I hope they are talking about the transmitter/receiver combo and not about their servos. Of course a slow servo would do no good to an otherwise fast system. But also note that 0.06 (airtronics) or 0.07 (Futaba if I remember correctly) is only very little difference, and this is the time needed by the servo for a FULL 60˚ sweep from one to the other end. Typically you use only a fraction of that distance and that should take only a fraction of the time.

As your example also indicates, I'm sure you can 'feel' differences of a few milliseconds. As an expert in (some areas of) visual perception I know that latencies in orders of 10ms can influence your performance. It's not that things get impossible but just that you need longer training to get them perfect. To give an extreme example, on some old organs there is quite some delay between the moment you press the key and the appearance of the sound. So a player may be effectively playing several notes in advance in fast pieces. Nevertheless some players get a flawless performance out of such crappy organs!


DavidH
I guess QPCM stands for Quadruple Pulse Code Modulation. Normal PCM uses TWO frequencies, one freq represents a 0 and the other a 1. QPCM uses FOUR frequencies, representing a 0,1,2,and 3 respectively. This means they can transmit 2 bits in parallel. Ideally that would thus double the transfer rate that MarkF mentioned from 8 to 16 kbit/s. I'm pretty sure that if they say QPCM that they will NOT use PPM. Maybe you can select the 'old' PPM modulation as a second transmission mode, but then you will need to get new receiver.
10-02-2003 Over year old.
HOMEPAGE  
 
 
w.pasman
Elite Veteran
Location: Netherlands

Ah I found the source of the HiTec performance

http://www.hitecrcd.com/Radios/eclipseQPCM.htm

No it doesn't say anything about the technology... except that they call QPCM "Quick"-PCM and not Quadruple as I thought...
10-02-2003 Over year old.
HOMEPAGE  
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Hi, W.Pasman!

For all the data to be consistent (other than Airtronics' nutty 200 uS claim), Hitec may well already have moved to a differential encoding scheme, and may have a frame rate more like 5.5 mS. That'd put Airtronics at perhaps 10mS, which would then jive with the Futaba and Graupner results (thanks for those figures)! This could then deliver a "response time" of 11 mS as measured from stick input to data ready inside the receiver prior to the next servo frame update (which I realize is perhaps meaningless, but since Hitec wants to interoperate with standard servos, they can't control that).

The (marketing) quote for the Hitec comes from the Eclipse PCM website . According to Hitec, Q-PCM actually just stands for "Quick" PCM (though it's possible that they've gone to a 4-level modulation scheme, I doubt that they've done so, since differential encoding is so much simpler and cheaper). [Sorry for the duplicate info - we doubled]

This is turning into a fun thread - thanks a bunch!

Cheers!
MarkF
10-02-2003 Over year old.
HOMEPAGE  
 
 
rckrzy1
Elite Veteran
Location: Hurst Texas

Who cares

So the 3 guys in all the world fly them can be happy knowing. Really .



Wildcat Fuels
10-02-2003 Over year old.
 
 
RLO1
Senior Heliman
Location: Chandler, AZ

I agree

Last I checked my fat thumbs were stilll only moving about 1000ms.

Seriously, I do not care how fast the equipment is. Todays servos/gyros/rx's are plenty fast enough for 99% of us out there.

What we need is power (at low amps), precision and reliability.

Thats why I fly JR digital servos, PCM rx and the futaba 401/9253 combo.

My two cents
-Ron
10-02-2003 Over year old.
HOMEPAGE  
 
 
JKos
Elite Veteran
Location: City of California in the state of Maryland

> So the 3 guys in all the world fly them can be happy knowing. Really

I'm sorry you feel that way. Some of us "geekier" folks like to delve into the finer points of our beloved hobby.

Some day our equipment will use spread spectrum and we will no longer have to worry about frequency control boards or radio impounds.

> I'll have to borrow an Eclipse and throw it on a digital scope...

I would if I had one of the newer QPCM versions but I bought Eclipse 7 about a week before they came out.

I think it would be very interesting to compare these different radios on a scope. I was thinking we could use the trainer input port as a controlled input and then watch the servo commands at the receiver. I believe data is still passed through the trainer port in PPM format. Is that correct?

Thanks,
John
10-02-2003 Over year old.
 
 
Tyler
Key Veteran
Location: Chicagoland area

Brian Tucker at Airtronics recently told me the advertised higher rate of transfer has nothing to do with the transmitter. The advantage is only seen in the receiver's technology. Simply put, " the receiver gets the signal to the servos faster than any other receiver on the market instead of the transmitter getting the signal to the receiver faster." The quick digital servos that Alan (and Panos and I) are now using then help make the heli respond quicker, too.

I also am using a 5.1v regulator to maintain constant battery voltage to my servos and rx. Off to test fly those new servos now. Will have to wait a little longer to decide if the Airtronics Stylus is worth upgrading from a pcm JR receiver and 8103 tranny. The jurry is still out on this one.

Tyler

Airtronics, YS Power, Switchglo ignitors, Jersey Modeler filled with Magnum
10-02-2003 Over year old.
 
 
MarkF
Senior Heliman
Location: Palo Alto, CA

Howdy!

rckrzy1: Thanks for another positive, affirming post!

To those who aren't interested in the issues we're discussing, that's OK with me. However, you would not have the reliability, speed, or precision that you have with today's equipment unless some folks actually did care, and attempted to make things better. I assume that you wouldn't be satisfied with a single-channel rubber-powered escapement radio in your Fury?

Engineering is all about continuous refinement, so that we all get better products. For me personally, the process of understanding how things work often leads to new ideas or solutions (with 26 patents so far). If this ultimately winds up making things better for R/C pilots, is that a bad thing? I do agree, however, that this thread might best be relocated to the Radio section, since it has gotten a little specialized.

Tyler: Thanks for the useful feedback! That is definitely the easiest way to go - apparently, Airtronix has simply upped the output frame rate to more than the conventional 50 Hz. If that's the case, their PCM receiver will only work with fast frame-rate servos. That certainly makes sense, and could be applied in addition to the other techniques mentioned above.

John: That's a great idea! Unfortunately, I believe that it's necessary with most PCM rigs to actually run in PPM transmit mode in order to use the buddy box cable - does anyone know of any exceptions to this? As an alternative, it wouldn't be too tough to jury-rig a PIC microcontroller chip to drive a servo that moves the stick, and use its on-chip A/D to sample the servo's position potentiometer to eliminate the uncertainty in the servo drive timing, and then time the pulses coming from the receiver's output. Heck, to make this even easier, you could just unsolder the wires from the TX pot, use a PWM output from the PIC to directly drive the transmitter (through an RC filter), then read back the receiver's servo drive pulses. Then, we could cut through the marketing fog and make direct comparisons between systems.

To those of you who enjoy learning more, thanks a bunch for the great posts!

Cheers!
MarkF
10-03-2003 Over year old.
HOMEPAGE  
 
 
5 pages [ <<    <    ( 1 )     2      3     NEXT    >> ]8353 viewsPOST REPLY
Thunder Power RC . Mikado Modellhubschrauber . GrandRC

.
.
Main Discussion > airtronics "worlds fastest"??
 PRINT TOPIC Advertisers 

Subscribe to This Topic

Tuesday, December 2 - 7:33 am - Copyright © 2000 - 2008 runryder.com | email | link to rr | runryder needs cookie