Black 5 smoke unit

[What About The Bee]

Hi 96RAF

We are in agreement. The reverser will affect the chuff because it affects cutoff.

There are 4 events for one steam port. Start of steam admission, end of steam admission (cutoff), start of exhaust and start of piston return.

To my understanding, start of steam admission begins at top dead center or thereabouts. Steam admission can be advanced or delayed by the reverser, but the degree of adjustment will be relatively small in relationship to piston stroke.

Steam admission ends at cutoff. The reverser affects the amplitude of the slide valve stroke. Depending upon the rate of flow of steam (regulator) and the time of cutoff, a certain pressure is achieved. That is, from start of steam admission, to end of cutoff, steam flows from one pressure vessel to the next. The duration of this period and the regulator constriction define how much steam, or pressure achieved in the destination vessel. The pressure achieved will affect the amplitude of the chuff.

The reverser does change the position of the slide valve with respect to the motion, primarily affecting the direction of the locomotive and cutoff. It does not change the relationship of the piston and wheels. Steam will be admitted at top of stroke and exhausted at bottom of stoke of piston.

The reverser will absolutely affect the sound. Solid agreement.

Where I think further clarity may be needed is the relationship of when steam exhaust occurs. That is the chuff and therefore important. Steam will be exhausted before the bottom of stroke of the piston. Examined logically, it cannot be after bottom, as the steam would then be compressed by the returning piston, decidedly inefficient! It will not be at mid expansion stroke, the steam will still be driving the piston forward, releasing the steam then would be wasteful. Indeed, I think it evident that for maximum efficiency, exhaust will occur as near to the end of stoke to permit enough time for the cylinder to reduce pressure, so as to not affect the return stroke.

Yet as the piston has mechanical linkage to the wheels, and it has been demonstrated that the exhaust is timed to the piston, it should be clear that the relationship of a chuff to a wheel orientation is relatively fixed. Exhaust occurs at end of piston stroke. Pistons are linked to wheels.

In my view, nearly all of this is icing on the cake. I do understand that others may perceive this differently. The reverser and regulator settings affect the sound only in degree. In my view, the fundamental issue is simply getting the correct number of chuffs and to have them occur at the relatively correct position of the wheels. Failing to get the fundamental correct makes the rest of it irrelevant. If you haven't got the correct number of chuffs for the velocity right, then of what use are the other modifiers?

Bee

[atom3624]

Hallelujah!

Let's hope Hornby think the same way ....

Al.

[96RAF]

Spot on Bee.

At present it has been regarded as essential to get the chuffs right within visual observation, but I have noticed with some profiles above visual perception the chuff can still sound blatantly wrong for various higher speeds. This throws out the plan to revert to simple chuff banding above a certain speed step. The complex sound curve remap helps here of course.

[What About The Bee]

I recognize this is a long standing topic and many are quite passionate about it.  

I do hope that this discussion is treated as intellectual discourse with truth being the goal. The truth will be found in the actual mechanism. I have no wish to upset other correspondents or ruffle feathers. Just to find truth.

Bee

[atom3624]

Never seen a better video explanation of the way, duration and intensity that the chuffs change with load and speed.

Many thanks to Nige for the fantastic video:

[DarkRedCape]

Well the R3991SS Flying Scotsman is now a Spring 2024 release, so I guess we won’t be getting a look at the gubbins any time soon after all.

[96RAF]

Once upon a time The Engine Shed told the tale of such new development but that news-desk hasn’t broadcast for a long time.

[DarkRedCape]

I know he’s not that popular around here, but I was just watching Sam’s review of a new P2 class and when he showed the axles, there was an unusual looking cog on one of them. Could this be part of the sensor for the smoke system? In car manufacturing, it’s sometimes cheaper to just fit a part to all models, even if that part isn’t used or needed, could be something similar here considering there is a smoke generating P2 model coming out.

Others will know far better than me. If you aren’t interested in the video, just skip to 21:40.

[atom3624]

I generally like most of Sam's reviews - he points you in the right direction, and of things to look out for as well.

That little 'daisy wheel' has got to be related to the 'chuff synching' one would expect.

Looks absolutely lovely.

Price? If that's where the market is then that's what it is!

Al.

[DarkRedCape]

I had the Earl Marischal and Lord President pre-ordered before the Rewards Points system was implemented, so I got both for £230 each.

Another thing to note about Sam’s video is at 22:15 it shows the chassis and there seems to be a large section at the front that would create a spacious void inside the body. I’m betting this is where the smoke generator lives on the SS version. Could be why the loco is so light, according to Sam’s weigh in.

[What About The Bee]

Hi DRC

That sensor Sam was clueless about appears very much to be a rotary encoder plate.

And I absolutely agree with your assessment, it is a tool to provide synchronization. The solution begins with a rotary encoder. The following writeup is my theory. Please note I have no insider information.

For those who do not know, a typical incremental rotary encoder has three channels. Sine, Cosine and Home. Sine and Cosine indicate direction of travel and with careful counting, where in travel. If the counting is measured against time, Sine and Cosine indicate velocity. The Home signal indicates a once per revolution reference.

Begin with Home. A CV in the decoder will indicate the delta between the arbitrary location on the rotary encoder and where top dead center of one wheel is. That is, the encoder need not be phased mechanically to the mechanism. Just plunk it on. We will install a programmable offset to adjust for the delta.

Another CV says how many chuffs per one revolution. Naturally, this is fixed. Divide the number of counts per encoder revolution by the number of chuffs. Thus the decoder knows exactly where each chuff is supposed to be, relative to rotary encoder position AND because the Home signal ties the rotary encoder to the valve gear, where each chuff is phased in relationship to the valve gear. This works for 4, 6, 8 or indeed any number of chuffs per revolution, software defined. There is nothing limiting you other than encoder resolution. You could even program crazy values, like 3.1415 chuffs per revolution!

Garrets and Big Boys may require 2 rotary encoders, as the timing between the two sets of drivers is not fixed.  There are two sets of 4 chuffs, but the phase relationship between those two sets can change, wheel slip being a primary example.

With the velocity read from the rotary encoder, we know the duration of chuff. Running fast, short chuff. Running slow? Long chuff. Etc. So play the sound of one chuff at each chuff event, with the pointer increment into that sound file adjusted by the velocity. A sound file is a sequence of bits. Move the sequence pointer as a function of velocity.

So what about the regulator and the reverser? These effectively change the amplitude of the chuff. It could be a table of CVs that indicate for a given velocity, what the engineman would normal have these controls.

For arbitrary control, the decoder could receive inputs that declare what the enthusiast has the controls at and modifies the sound accordingly. That might just be a bridge too far for current technology. Feasible? Yes. The enthusiast would need a way to command these values, instead of velocity, the enthusiast would command regulator and reverser. It would put the enthusiast "into the cab". The enthusiast would drive just as the engineman would. The sound would be correct!

This works for a vast array of locomotives, making this near universal. The only thing to change would be the sound file of one chuff, per locomotive.  

Bee

[Tolak]

Interesting that your description has the start of the inlet and the start of the exhaust at TDC.

I though the reverser/cutoff effectively reduced the effect of the inlet valve (effectively the valve throw, and hence the "gain" of the inlet valve), but would not have changed the timing.

Hence if the inlet valve at full opening opened at TDC and shut at TDC+70° (ie resulting in a 70° opening period), backing off the valve opening would affect both ends of the opening timing, and hence might achieve opening at TDC+20°, and closing at TDC+(70-20)°, so the valve opening would be shut-off by 40° less period (ie only 30° opening period in this case).

This means the exhaust would change timing, too, by a similar effect. I suspect it would take video footage to see that when the cut-off would be advanced, since the timing would be hard to see when the loco was running fast enough to benefit from the change of the cut-off.

And to that end, I agree that the chuff timing at lower speed is the main definition; having the back-EMF (error) define the aggression of the chuff might provide a "simple" mechanism, where higher error (ie working harder) gives a more agressive chuff, and low error (eg coasting) would give a softer chuff.

I also note your comment about the rotary encoder; this is much more detailed than our most common rotary machine (the car engine) that only has a TDC sensor and evaluates the timing of the other events (sparks) based on % of duration between the TDC events. With both systems, the TDC (Home) event has to be remembered, so loss of power would lose sync of the chuffs until the next TDC/Home event occurs.

Thanks for the discussion,

Regards,

[What About The Bee]

Hi Tolak

Every decoder on the market today has non-volatile memory on board, preserving CV settings. No reason at all why one of them could not be the offset from home pulse to TDC. Any further adjustments of the chuff timing whilst operating be handled programmatically, for example, the reverser.

That disk displayed in Sam's video has all the characteristics of a simple rotary encoder. That is, uniform blanks and openings around the circumference of the disk. The encoder reader head, not present in the video, will have two channels, phased 90° to each other, with the corresponding equivalent grating to the disk. The technology is fairly simple and well known for at least ½ century. In consideration that the encoder resolution need not be in the millions of counts per revolution (I dare say 256 counts per revolution would be more than adequate), the expense of such an encoder will be low.

Just my comments from the peanut gallery.

Bee

[96RAF]

The cogged disk has been described on another forum as a reluctor disk which would tend to have equally spaced micro-magnets, typically triggering a Hall effect sensor. A clogged disk would tend to trigger an IR beam.

The decoder by way of CVs will likely be able to set length of chuff and intensity of background as the chuff does not ever stop passing steam it just varies in force.

Excuse my layman’s description as I an not a railwayman, aircraft are my forte.

[Tolak]

Hi, Bee, thanks for the update.

I have just watched Sam's review, and it appears that the vanes on the disc have six elements, so I would expect these use an optical sensor to drive the chuff directly, and not to require any more sophisticated decoding. (For the Black 5, we would need 4 vanes per rev, but the FS can use the same 6 vane disc. Has anyone done a similar review of the new Black 5 to see what its vane looks like?)

After all, we only really need the six beats per revolution; any loss of sync *could* be corrected by the decoder, but would probably be ignored. If the sync is so far out that you can tell that the timing of one set of valves can be seen to be incorrect, then I suspect we would have to force the vane round to correct that rotational error (effectively re-"quartering" the vane. But the error can only be 60°, so the perceived error will be ±30°, which might not be enough to notice. At least the pulse rate will be correct!

And Interesting to see what the pulse timing will be in reverse!

Regards, Tolak.

[What About The Bee]

I think we can agree that it is an encoder of some type, the only disagreement is the method by which the encoding is resolved.

My guess was a true optical encoder, complete with lissajous that can be further resolved with Sine and Cosine. It is just that, however, a guess. I have no insider knowledge whatsoever!

Tolak has an optical trigger. I would counter that the system could not tell which of the 6 vanes would be "the one". It could be made to function that way, indeed, that is essentially the Lionel solution. One pulse per revolution informs the entire chuff system.

96RAF has reported a magnetic ring. Magnetic encoders are a well accepted encoding solution. For this, I would question the need for optical vanes whatsoever, a simple disk with magnets glued on would suffice.

Yet all of these, my guess included, are variations on a theme. If you want the chuffs to come out right, you need orientation of phase.

I am happy that Hornby appear to be making strides in this direction!

Bee

[96RAF]

@Bee

… whenever I get solid info about how this ASU system works then I will post it, but as of now I am unaware except as already informed. It will however to date be a hybrid in that slow chuffing will be synchronised by such a method described but above visual discernment it will revert to batch chuffing.

[Tolak]

@Bee

i don’t think the rotary encoder needs to know which pulse is “the one”, as it only needs a trigger to drive the next chuff. Hopefully, the chuff sequence is matched to the loco, ie six chuffs, but not all the same.

Looking at Sam’s strip down, I would estimate that the vanes are 40 degrees wide, so if the chuffs were exactly sync’d going forward, they would be approx 20 degrees late in reverse. (Note, this is the back edge of the adjacent vane when running in reverse, so the gap accounts for the delay). Still chuffing at 6 beats to the bar, but with slight worse sync.

if this comes from the TXS decoder, I wonder if that will become a useful add-on to all these sound chips? I would expect the vane to be quite easy to fit; easier than the magnets! (Or, using the other function of an optical decoder, use a reflective ring on a wheel to trigger the chuffs)

As RAF says, we can look forward to seeing this in the flesh. Or at least, the Rohss friendly solder.

Anon

[Tolak]

@Bee

”Garrets and Big Boys may require 2 rotary encoders, as the timing between the two sets of drivers is not fixed.  There are two sets of 4 chuffs, but the phase relationship between those two sets can change, wheel slip being a primary example.”

For the Rivarossi Big Boy, both wheel sets are driven from the same motor, so only one rotary encoder might be needed. Unless, like me, you have stripped it down to ensure that the valve gears are not “in step”.

;)

[atom3624]

I've often thought of this one - Big Boy and that they should never be considered to be permanently in synch ..

Al.

[What About The Bee]

Hi Atom

Agreed, and they are not! I would suggest that they are almost never in synchronous timing.

The two sets of valve gear are independent on the prototypes. There is no guarantee that driving set one matches driving set two. Maybe on a model, but frankly, that isn't prototypical.

Bee

[What About The Bee]

Hi Tolak

Diplomatically, the name of this thread is "Black Five smoke unit". Forgive me if I am wrong, as that locomotive isn't in my era, but I thought it only had 2 cylinders. I could be wrong, but ....

So 4 chuffs, 90° apart.

Consider the disk with 8 vanes. Rising edge triggered, you get 8 triggers per rotation. Falling edge triggered, you get 8 triggers per rotation. Level triggered will require a low pass filter to inhibit a stuttering trigger.* But even that will be 8 triggers per rotation.

Of course, we can skip every other trigger, getting back to 4 triggers per rotation.

But if you pick the wrong edge on power up, you can be 45° out of phase with the valve gear. So you rotate the disk relative to the shaft to align it. Power down, and back up. Did it pick the right edge?

I look forward to seeing how Hornby do this. If the phase relationship is only good to 45°, however, customer dissatisfaction will be very evident.

Bee

*watch out for higher speeds, such that the low pass filter doesn't filter out required triggers.

[96RAF]

So then we need another disk for 3 cylinder locos. A disk with asymmetric mark/space ‘blades’ maybe.

[Going Spare]

And for the Lord Nelson with 8 beats to the revolution?

[What About The Bee]

Perhaps now you may see why I have proposed an encoder. The number of vanes does not correspond to the number of chuffs per revolution for an encoder. Rather, it serves to divide up the disk into parts, termed "counts". Once the reference is found, we can can keep counting up or down as each count occurs. Thus the cuff occurs on a specific count total.

But Bee, I hear you say, if a count is a vane, it is the same thing, isn't it? Nay Nay!

Put a light receiver on one side of the vanes, an emitter of light on the other side. The vanes alternatively block and permit light. Properly grated, the receiver will detect a sinusoidal light level. Put two emitters, and shift one such that it is 90° out of phase to the other. Now you have Sine and Cosine! With Sine and Cosine, you have a circle (lissajous) for each vane. Divide that circle or lissajous mathematically into 8 parts (45°) and now you have 8 counts per vane or 64 counts per full disk revolution.

You may see that I can divide the lissajous into any number of arbitrary counts. The multiplier we used, back in the 1980s, for our own self designed encodering systems was 256, or 256 counts per lissajous. 8 vanes then yields 2048 counts per one disk revolution.

Assign a chuff to occur at counts 512, 1024, 1636 and 2048. Viola!

Works for any number of chuffs. 6 chuffs per disk revolution? Divide 2048 counts by 6. 8 chuffs per revolution? Divide 2048 by 8, yielding a chuff every 256 counts.

Therefore, one disk works for every/any locomotive, independent of the number of chuffs per encoder disk revolution. Its done mathematically, not a direct trigger.

I think the easiest way to see this effect is to put the palms of your hands together, fingers spread. Rotate one palm against the other, observe the light coming thru your fingers. We can sample that light, and determine its intensity. That gets you Sine, as per above.

Again, I have no insider knowledge and could be absolutely incorrect. This is just my guess. An educated guess, based on experience, but still a guess.

Bee