Futaba 4PX
#2266
Tech Apprentice
Interesting data and awesome work! Thanks!
So frame rate is 3.3 but this is already past useful because the servo can't update that fast.
Min latency is 2.2ms with average being 5.3ms.
We still have no idea what the M12/M12-S specs are with you test. We also have no idea if Sanwa's stated values are for minimum or average latency. Do we even know if their specs were tested at the component or the servo level?
It would be neat to see the following two things:
1) specs for M12S with your tests
2) spec of 4pX with a 614 Fasst Rx ( I assume it automatically links on C2 mode)
So frame rate is 3.3 but this is already past useful because the servo can't update that fast.
Min latency is 2.2ms with average being 5.3ms.
We still have no idea what the M12/M12-S specs are with you test. We also have no idea if Sanwa's stated values are for minimum or average latency. Do we even know if their specs were tested at the component or the servo level?
It would be neat to see the following two things:
1) specs for M12S with your tests
2) spec of 4pX with a 614 Fasst Rx ( I assume it automatically links on C2 mode)
I am meting a friend on Saturday with the little older M12 and I will get the chance to test it to compare. Also a friend with a MT4S.
I will borrow a 614 fasst rx to do direct compare with 4PK, and also create histogram of the 4PK and the sanwa radios.
#2267
Tech Fanatic
iTrader: (2)
The RX and 4PX I tested had 3.0ms servo update and not 3.3ms as the 4PK. So a little increase in the Hz. And digital servos can handle the update rate.
I am meting a friend on Saturday with the little older M12 and I will get the chance to test it to compare. Also a friend with a MT4S.
I will borrow a 614 fasst rx to do direct compare with 4PK, and also create histogram of the 4PK and the sanwa radios.
I am meting a friend on Saturday with the little older M12 and I will get the chance to test it to compare. Also a friend with a MT4S.
I will borrow a 614 fasst rx to do direct compare with 4PK, and also create histogram of the 4PK and the sanwa radios.
#2268
Tech Apprentice
The 4px with R614FS FASST Receiver in green and the T-FHSS R304SB in blue.
The Red and Blue lines are standard deviations, high and narrow is better.
On average the FASST is slower, but it doesn't have the same pattern on the few unusual high latency hits, they are there a few very high latency spikes but not the same pattern as with the T-FHSS. If we only look at the compact area where 80% of the samples are the difference is 1ms but with all the bad random high samples that are very frequent with the T-FHSS the average difference is only 0.5ms.
Also the FASST is 303Hz update rate and the T-FHSS is 333Hz.
The Red and Blue lines are standard deviations, high and narrow is better.
On average the FASST is slower, but it doesn't have the same pattern on the few unusual high latency hits, they are there a few very high latency spikes but not the same pattern as with the T-FHSS. If we only look at the compact area where 80% of the samples are the difference is 1ms but with all the bad random high samples that are very frequent with the T-FHSS the average difference is only 0.5ms.
Also the FASST is 303Hz update rate and the T-FHSS is 333Hz.
#2269
Tech Adept
Thank you, this is very valuable information.
#2270
Tech Apprentice
And the question many people have asked. Is the Sanwa M12 any faster as they claim?
I have the answer. And it is.... drum roll.... yes!
Sanwa M12 (not M12s) in Green, And 4PX with included T-FHSS reciever in Blue.
Avg delay is 4.080ms on M12 and 4.636ms on 4PX. The random bad values on the Futaba is really hurting the numbers. But if you look at where the "wall" begins the Sanwa have 0.3ms advantage from just being faster somewhere in the radio communication chain. And the Sanwa have a servo update of 384Hz making the distribution narrow compared to 333Hz on the 4PX. The Sanwa also have very very few random high samples.
Sanwa was tested in the high speed mode for normal servos. I did a read with my logic bench too see how fast it was with the Sanwa special servo communication. To my suprise the update rate was the same at 384Hz and it used ~130-470µs pulses with 300µs as center instead of 1000-2000µs in normal servos. I did not have time to reprogram my testprogram and make a sample with this mode but my guess is that it will have the same outcome with a 1.5ms shift due to shorter communication signal to the servo but since the Hz is the same the spread in the standard deviation will not be better.
The second finding we did was that when we changed from normal analog mode to the faster high speed or digital mode on both 4PX and the M12 a rebind of the receiver was necessary for the RX to use the new speeds. Something I guess 90% of people have changed without rebinding and think they use the faster mode but they are not. On 4PX the digital servo mode is default and should not harm any other than analog servos that will burn up because you change the setting and the radio will not change it until you rebind. On the M12 the normal slower mode is default and changing it to the faster with out rebind will have no effect and it will use 96Hz. And 96Hz sound a bit to high to be analog compatible if any people have any issues with analog servos this might be a answer to that. Other radios have about 60Hz in Analog/normal mode.
ps
Thank you very much Kristian who let me test his radio!
EDIT: Updated Futaba results with telemetry off. M12 was tested with No telemetry functions to be fair.
I have the answer. And it is.... drum roll.... yes!
Sanwa M12 (not M12s) in Green, And 4PX with included T-FHSS reciever in Blue.
Avg delay is 4.080ms on M12 and 4.636ms on 4PX. The random bad values on the Futaba is really hurting the numbers. But if you look at where the "wall" begins the Sanwa have 0.3ms advantage from just being faster somewhere in the radio communication chain. And the Sanwa have a servo update of 384Hz making the distribution narrow compared to 333Hz on the 4PX. The Sanwa also have very very few random high samples.
Sanwa was tested in the high speed mode for normal servos. I did a read with my logic bench too see how fast it was with the Sanwa special servo communication. To my suprise the update rate was the same at 384Hz and it used ~130-470µs pulses with 300µs as center instead of 1000-2000µs in normal servos. I did not have time to reprogram my testprogram and make a sample with this mode but my guess is that it will have the same outcome with a 1.5ms shift due to shorter communication signal to the servo but since the Hz is the same the spread in the standard deviation will not be better.
The second finding we did was that when we changed from normal analog mode to the faster high speed or digital mode on both 4PX and the M12 a rebind of the receiver was necessary for the RX to use the new speeds. Something I guess 90% of people have changed without rebinding and think they use the faster mode but they are not. On 4PX the digital servo mode is default and should not harm any other than analog servos that will burn up because you change the setting and the radio will not change it until you rebind. On the M12 the normal slower mode is default and changing it to the faster with out rebind will have no effect and it will use 96Hz. And 96Hz sound a bit to high to be analog compatible if any people have any issues with analog servos this might be a answer to that. Other radios have about 60Hz in Analog/normal mode.
ps
Thank you very much Kristian who let me test his radio!
EDIT: Updated Futaba results with telemetry off. M12 was tested with No telemetry functions to be fair.
Last edited by condac; 07-09-2017 at 01:15 AM. Reason: No telemetry
#2271
Tech Apprentice
I almost forgot. I have tested the 4PX with a S-FHSS reciever R204GF-E
Avg latency is 10.1ms. almost 5ms slower than T-FHSS. And S-FHSS is running the servo update at 147Hz.
Here is S-FHSS with R204GF-E RX in Green and T-FHSS R304SB RX in Blue:
Avg latency is 10.1ms. almost 5ms slower than T-FHSS. And S-FHSS is running the servo update at 147Hz.
Here is S-FHSS with R204GF-E RX in Green and T-FHSS R304SB RX in Blue:
#2272
Tech Fanatic
iTrader: (2)
And the Sanwa M12 is just a breath faster than the 4pX with TFhss, but it also has better average latency due to its maxed out frame rate of 384hz
#2273
Tech Apprentice
Awesome!!!!! Now we know that the Futaba techs are just as clueless about their products. Lol. So the Fasst rx has more latency, lower frame rate, but higher resolution? The TFhss has less latency, a higher frame rate, less resolution, and drops more packets (thereby increasing its average latency)??
And the Sanwa M12 is just a breath faster than the 4pX with TFhss, but it also has better average latency due to its maxed out frame rate of 384hz
And the Sanwa M12 is just a breath faster than the 4pX with TFhss, but it also has better average latency due to its maxed out frame rate of 384hz
#2274
Tech Adept
Have you tested 4PX after latest firmware update with racing mode ON?
#2275
Tech Fanatic
iTrader: (2)
I assume that even though the minimum latency of the M12 vs the 4PX are virtually the same, the M12's better average latency, along with it's better frame rate makes it feel a little more connected. The M12s should be even better.
It will be interesting to see if a firmware update to the 4pX for the use of the new SR receiver will see a boost in frame rate, and reduction in latency?
Can you test an M12-S? After experiencing how much easier it was to drive my car with the T-Fhss Rx, I can see why people like the M12/MT4s/M12-S lineup. Futaba has some work to do.
With that said, it does look like the latency average and frame rate issue should be relatively easy to fix. Hopefully for us 4pX owners a firmware update to the new SR Rx spec will move us past the M12 at least.
#2276
Tech Apprentice
#2277
Tech Adept
#2278
Tech Apprentice
Firmware update from 1.2 to 1.4. No change.
Race mode on or off ~0.1ms change I call that No change.
Telemetry off..... Yes big change. Avg latency went from 5.165ms to 4.636ms. And it impacted the strange areas on the histograms making the std deviation much better. So Futabas claim that telemetry have no impact on response is false, but the impact is probably not noticeable because we talk about <1ms here.
Green telemetry on, Blue Telemetry off.
I have also updated the M12 comparison picture in previous post to compare both systems without telemetry.
Race mode on or off ~0.1ms change I call that No change.
Telemetry off..... Yes big change. Avg latency went from 5.165ms to 4.636ms. And it impacted the strange areas on the histograms making the std deviation much better. So Futabas claim that telemetry have no impact on response is false, but the impact is probably not noticeable because we talk about <1ms here.
Green telemetry on, Blue Telemetry off.
I have also updated the M12 comparison picture in previous post to compare both systems without telemetry.
Last edited by condac; 07-09-2017 at 01:20 AM.
#2279
Tech Adept
Great! Thank you for you effort and published information.
#2280
Tech Fanatic
iTrader: (2)
Beautiful work Condac.
Thank you for clearing up all of the murky data that Futaba was giving.
So in summary thus far we have:
-4PX is roughly equal to the M12.
-4PX Minimum latency stilled turned out to be around 2.2ms - 2.1ms (race mode); even though it was calculated based upon Futaba's unreliable data.
That's still better than some of Futaba's data that was originally pointing towards 2.7ms - 2.3ms.
-M12 still has a slightly better Average Latency spec (4.1ms vs 4.6ms)
-M12 uses the highest theoretical frame rate of 384hz vs 4PX of 333hz which may be helping to tighten Average Latency specs.
-M12-S & MT4-S will be faster than 4PX in both latency and frame rate
(Any way we can get you to test both of these Condac?)
-Fasst Rx is faster than SFhss both in Frame Rate and Latency.
-TFhss (big surprise) is faster than Fasst in Frame Rate and Latency.
-Telemetry hurts Average Latency (at least for Futaba).
-4PX Frame Rate and Latency are affected by the Rx that is used.
-Rx Frame rate and latency are more key for the 4PX than Rx resolution.
-Frame rate has a similar impact upon servo control (connectedness) as resolution has been debated to have.
This is why we test: Based upon this independent test data I was able to test a different Rx and found a substantive difference on the track with car control.
For me, car control = "connectedness", instead of "responsiveness" alone (although I do think that it IS important; which is why I have everything maxed out on my 4PX ).
We wouldn't have known any of this for sure Condac without you. Thanks again. Now, get a hold of an M12-S, MT4-S, and the new 7PX when it comes out, and show us the data.
Thank you for clearing up all of the murky data that Futaba was giving.
So in summary thus far we have:
-4PX is roughly equal to the M12.
-4PX Minimum latency stilled turned out to be around 2.2ms - 2.1ms (race mode); even though it was calculated based upon Futaba's unreliable data.
That's still better than some of Futaba's data that was originally pointing towards 2.7ms - 2.3ms.
-M12 still has a slightly better Average Latency spec (4.1ms vs 4.6ms)
-M12 uses the highest theoretical frame rate of 384hz vs 4PX of 333hz which may be helping to tighten Average Latency specs.
-M12-S & MT4-S will be faster than 4PX in both latency and frame rate
(Any way we can get you to test both of these Condac?)
-Fasst Rx is faster than SFhss both in Frame Rate and Latency.
-TFhss (big surprise) is faster than Fasst in Frame Rate and Latency.
-Telemetry hurts Average Latency (at least for Futaba).
-4PX Frame Rate and Latency are affected by the Rx that is used.
-Rx Frame rate and latency are more key for the 4PX than Rx resolution.
-Frame rate has a similar impact upon servo control (connectedness) as resolution has been debated to have.
This is why we test: Based upon this independent test data I was able to test a different Rx and found a substantive difference on the track with car control.
For me, car control = "connectedness", instead of "responsiveness" alone (although I do think that it IS important; which is why I have everything maxed out on my 4PX ).
We wouldn't have known any of this for sure Condac without you. Thanks again. Now, get a hold of an M12-S, MT4-S, and the new 7PX when it comes out, and show us the data.