SWAY - a synth modeled after the Yamaha SY77

[IvyBirds]

Yes, the feedback loop in 3 is attached to itself, and I believe the diagram shows that too. If it were attached to 1 you would see a wire going all the way from the bottom output of 1 back to 3. You can see the output of 1 is not connected to the input of 3. But the output of 3 is both connected to the input of 3 and the input of 1 and 2. So 3 modulates 1 and 2 (and itself through feedback).

I never said that 1 was modulating 3. I have always said that 3 was modulating one via a feedback loop.

Internally the way 3 can modulate both 1 and 2 is indeed through the feedback bus

And that is not the way the DX7 does it as it only has one feedback loop which is being used to send feedback to itself

but that's equivalent to direct modulation or modulation through the accumulator because it's not actual feedback yet.

But it's fundamentally different than how the DX7 did it

The 3 feedback busses act as 3 additional storage units that you can use to modulate lower level operators (they are named FF1/FF2/FF3 in Sway). Only in the next sample frame do they get filtered and are used for actual feedback. It doesn't matter if you modulate a carrier by OP, ACC, or FF1/FF2/FF3, as long as they contain the same value then the modulation will be the same.

Awesome so it's not the same as in the DX7

I guess they had to connect it this way for algo 40 because the accumulator could not be used, as it needs to preserve the output of the 6-5-4 stack. In the DX7 that's not the case because the 3-2-1 stack there is calculated last so they can use the direct operator input instead. Not sure why they reversed the role of 6-5-4 and 3-2-1 between the DX7 and the SY77, but the output is mathematically equivalent if you swap the settings for 6-5-4 and 3-2-1. And the algorithm is configured this way out of the box, so I think the writer of the SOS article is simply mistaken.

So what you are saying is on the DX7 they are using direct operator input but on the SY they are using a Feedback Bus. So there is a fundamental difference between the two algorithms, they are not identical, which again has always been my point. While they accomplish the same goal they do so with different methods which requires a different workflow

You said the three feedback busses on Sway are called FF1/FF2/FF3. I will assume each of this busses has its own level control somewhere that is different from the operators level itself correct?

Currently there is no way to save custom algorithm presets, but that's an interesting idea! I'll add it to the list of possible future features.

Thanks it would be a very handy feature

[IvyBirds]

Same is on TG/SY by default as confirmed by gentleclockdivider.

This may have been added in a firmware update, so I wouldn't necessarily fully believe articles online which might have deprecated info. Do you have an actual TG/SY unit? That would be the easiest way to check.

Also, do you really think sheaf would do this incorrectly after spending years working on Sway? I don't think so.

Awesome but all I asked for was the numbers of what feedback loops or busses are being used.

And I would appreciate it if you would stop trying to put words into my mouth, I never suggested that sheaf did anything incorrectly, he made a SY77 Emulation not a DX7 Emulation and they are two different machines

From your posts you seem to lack a fundamental understanding of how the DX7 Algos work and are "wired" so to speak, and how that relates to how the SY77 Algos work and are wired. Rather than desperately trying to prove me wrong when you don't even have the hardware you should do some research. As sheaf just confirmed in the SY77 algorithm in question, Yamaha had to use a feedback bus so that a single modulator could modulate two carriers

There is a specific workflow issue with that approach which is why I simply asked what feedback loop numbers are being used so that the modulator in question can modulate itself and another operator acting as a carrier

Way back before you started your personal diatribe against me in some desperate attempt to prove me wrong I said the following

However Yamaha giveth and Yamaha Taketh Away and we also lost algorithms 19 and 20 where each of those have three modulators and two carriers both modulated by one modulator and there is no direct equivalent in SY77. One could argue that doesn't matter as you can add feedback loops going from a modulator to a carrier which can accomplish the same thing

You said that was wrong

However Yamaha giveth and Yamaha Taketh Away and we also lost algorithms 19 and 20 where each of those have three modulators and two carriers both modulated by one modulator and there is no direct equivalent in SY77. One could argue that doesn't matter as you can add feedback loops going from a modulator to a carrier which can accomplish the same thing which is true, but it's also not the same workflow wise which means the DX7 is able to do things the SY77 can't namely have Algorithms 19&20

Not correct. Algos 19-20 are algos 40-41 on SY.

Only I wasn't wrong I was correct as has been confirmed by sheaf.

If you would drop the personal vendetta against me you might just learn something about these incredible machines that you don't even own. So please since you have the Beta of Sway and claim these feedback loops exist, all I am asking is what feedback loops/busses are being used to accomplish that there are only 3 available, it's not really hard. I just don't have the Beta to be able to look it up myself

[EvilDragon]

There is no fundamental difference in practice, because those two implementations actually result in exactly the same sounds.

[sheaf]

So what you are saying is on the DX7 they are using direct operator input but on the SY they are using a Feedback Bus. So there is a fundamental difference between the two algorithms, they are not identical, which again has always been my point. While they accomplish the same goal they do so with different methods which requires a different workflow

No, that's not what I'm saying. Both the DX7 and the SY77 use the direct output of operator 3 as input for operator 1 & 2, regardless of where they store it in the meantime. The output is exactly the same, the computation is exactly the same. For all intents and purposes these algorithms seem identical to me, and right out of the box.

You said the three feedback busses on Sway are called FF1/FF2/FF3. I will assume each of this busses has its own level control somewhere that is different from the operators level itself correct?

No, FF1/2/3 don't have a level control. They're just additional registers to store an operator output.

[EvilDragon]

I hope we can put this to rest now. Algos are identical in practice, and with no extra work. Which is what I said and gentleclockdivider and sheaf confirmed.

[Uncle E]

Frankly, this is OT.

[EvilDragon]

Tangentially related super titillating FM discussion, what's not to love?!

[IvyBirds]

Subject: SWAY - a synth modeled after the Yamaha SY77

I hope we can put this to rest now. Algos are identical in practice, and with no extra work. Which is what I said and gentleclockdivider and sheaf confirmed.

Actually they are not identical in practice and it makes a difference

On the SY we only have 3 feedback loops. That is a finite number we don't get any more

Because the two Algos in question are not identical in practice we have to waste one of those three loops which means we we only have two left

You can continue to jump up and down all you want and pretend they are identical but they are simply not and in the process we lose the ability to have a feedback loop be used for a different purpose

So what you are saying is on the DX7 they are using direct operator input but on the SY they are using a Feedback Bus. So there is a fundamental difference between the two algorithms, they are not identical, which again has always been my point. While they accomplish the same goal they do so with different methods which requires a different workflow

No, that's not what I'm saying. Both the DX7 and the SY77 use the direct output of operator 3 as input for operator 1 & 2, regardless of where they store it in the meantime. The output is exactly the same, the computation is exactly the same. For all intents and purposes these algorithms seem identical to me, and right out of the box.

Except for the fact that they waste a Feedback Bus in the process. Essentially you are using a bus to send the direct output of operator 3 to operator 1. Which is silly because the direct output of operator 3 is already being sent to operator 2 without that bus.

This only leaves two of the three feedback busses available to the algorithm for other things

If they had implemented it like they did on the DX7 you will still have 3 different feedback busses available on the matrix to do different things and connect different operators together

So while yes you can do this and duplicate what is being on the DX7 you also lose 33% of the advantages AFM has over the DX7 when it comes to extra routings

In 1990 I already had multiple DX7s, I upgraded to a SY for extra functionality which got diminished because Yamaha didn't duplicate the Algos

[Uncle E]

OT

[EvilDragon]

FFS.

Classic goalpost moving.

Nevermind the fact that DX7 had just a single feedback path.

The bottom line is, you can get the same algorithm as on DX7, it will sound the same. Ivybirds is wrong, and we can move on.

[sheaf]

Subject: SWAY - a synth modeled after the Yamaha SY77
Except for the fact that they waste a Feedback Bus in the process. Essentially you are using a bus to send the direct output of operator 3 to operator 1. Which is silly because the direct output of operator 3 is already being sent to operator 2 without that bus.

This only leaves two of the three feedback busses available to the algorithm for other things

If they had implemented it like they did on the DX7 you will still have 3 different feedback busses available on the matrix to do different things and connect different operators together

So while yes you can do this and duplicate what is being on the DX7 you also lose 33% of the advantages AFM has over the DX7 when it comes to extra routings

In 1990 I already had multiple DX7s, I upgraded to a SY for extra functionality which got diminished because Yamaha didn't duplicate the Algos

I agree we should probably put this to rest now, but I feel like I must defend the honor of the SY77 designers.

Even if they connected it like the DX7 it would still have one feedback bus occupied with operator 3. Otherwise it could never have feedback!

Operator 2 doesn't need the bus because it can draw the operator 3 output from the OP register (which always contains the output of the previous operator). Operator 1 can't do that because there OP is filled with the output of operator 2. So it gets it from FF1 instead, which necessarily had to contain the operator 3 output anyway. Perfect!

[IvyBirds]

I agree we should probably put this to rest now, but I feel like I must defend the honor of the SY77 designers.

Even if they connected it like the DX7 it would still have one feedback bus occupied with operator 3. Otherwise it could never have feedback!

Operator 2 doesn't need the bus because it can draw the operator 3 output from the OP register (which always contains the output of the previous operator). Operator 1 can't do that because there OP is filled with the output of operator 2. So it gets it from FF1 instead, which necessarily had to contain the operator 3 output anyway. Perfect!

Sure let's put this to rest

This is ago 41 on the SY using the picture you uploaded

Screenshot_20250920-154117__03.jpg

In this example I want to set the three feedback lanes as drawn with the red lines

I want Operator 3 to modulate operator 1&2 and make 2 new waveforms Then have those be sent to operator 5 where those waveforms become modulators. The balance of those waveforms would be controlled via the levels of 1&2 either from their envelopes or from a controller since as you said Sway (and SY77) doesn't have a level control on the Feedback Busses

At the same time I want Operator 6 to send a feedback loop to itself.

How do I do this in Sway? If the Yamaha engineers had simply recreated the DX7 Algo it would be simple, but they didn't do that, instead a feedback loop got wasted which is the green line in the image and since we don't have 4 feedback lanes we can't

Looking forward to your response, this exact scenario has been bouncing around in my head for 35 years. It was my White Whale until I got some plugins that could pull it off

[IvyBirds]

FFS.

Classic goalpost moving.

Nevermind the fact that DX7 had just a single feedback path.

The bottom line is, you can get the same algorithm as on DX7, it will sound the same. Ivybirds is wrong, and we can move on.

Only I didn't move any posts you are. I have zero need or desire to recreate the DX7 on the SY77 engine and never said I did. Since I owned multiple DX7s when the SY came out I had no need to back in the day and since I currently own every DX7 plugin I am aware of, I have no need to do so now

That would be you lying about me and putting words in my mouth

I want to do this and things like this

Screenshot_20250920-154117__03.jpg

So please, since you claim I am wrong and you are right tell me how to connect the three feedback lanes on the SY77 engine. It should be easy since you say I am wrong and you are right

To do so I would need to disconnect the feedback lane going from 3 to 1 but then 3 is no longer modulating 1 which it is in the DX7 Algo you say is copied

So here is your chance go for it

Edit:And before you falsely accuse me once again if moving the goalposts remember you have been saying the two algorithms are identical, and I have been saying they are different

If they are identical as you claim tell me how to do it

[planetearth]

You must be fun at parties.

[sheaf]

Sure let's put this to rest

This is ago 41 on the SY using the picture you uploaded

Screenshot_20250920-154117__03.jpg

In this example I want to set the three feedback lanes as drawn with the red lines

I want Operator 3 to modulate operator 1&2 and make 2 new waveforms Then have those be sent to operator 5 where those waveforms become modulators. The balance of those waveforms would be controlled via the levels of 1&2 either from their envelopes or from a controller since as you said Sway (and SY77) doesn't have a level control on the Feedback Busses

At the same time I want Operator 6 to send a feedback loop to itself.

How do I do this in Sway? If the Yamaha engineers had simply recreated the DX7 Algo it would be simple, but they didn't do that, instead a feedback loop got wasted which is the green line in the image and since we don't have 4 feedback lanes we can't

Looking forward to your response, this exact scenario has been bouncing around in my head for 35 years. It was my White Whale until I got some plugins that could pull it off

But they recreated the DX7 algo, exactly. There just is no other way to do it.

You're asking for 4 feedback loops so that's not possible no matter how you slice it. Part of that particular algorithm is that operator 3 sends feedback to itself. So that's one feedback path fixed. Then you want feedback from 1 to 5, from 2 to 5 and from 6 to 6. That's 4 feedback paths in total. Now if you want to get rid of the fixed one from 3 to 3 so you have one more path available to configure as you wish, you can do that with a custom algorithm, but then we're no longer talking about DX7 algo 20.

The green line is not drawn correctly btw, and I think this is where the confusion starts. Feedback goes from an output to an input. Your line goes from an input to an input. In this case your green line should be a loop drawn like the one you did for 6. The diagram should be understood like this: the output of 3 is sent directly to 1, directly to 2, and as feedback to itself. The feedback doesn't involve 1.