Ayre wants $1.5K for DSD'ed QB-9

[wisnon]

At last, something for which I know the answer!

 

Between 1431km and 1708km.

 

About 22-25 hours driving time (including ferry)

 

Do I win one of those Lampizator DSD DACs? That would be great.

 

 

Sure, just as soon as I win Powerball BIG. LoL

[elcorso]

How far is Finland from Poland?

LoL

 

And from the Rainforest by tree vines?

 

I'm just finishing mine, completely analog and tubed, of course...!

 

 

Roch

[Charles Hansen]

This is music to my ears!! No pun intended

 

If I had it my way the Ayre "R" DAC would have:

 

-An analog volume control implementation, the same as in the KX-R preamp

-Balanced analog inputs to bypass additional analog preamp

-Dual-mono design

-At least 21 bits of resolution (>130 dB SNR)

-130 db of channel separation (20 Hz - 20 KHz) for rock solid imaging

-A nice built-in LCD screen that displays the artwork of the album being played

 

Is that too much to ask for?

 

Dear Silvertone, this is the fourth time that I have tried to answer. The forum message system keeps eating my reply.....

 

I will try one more time. I can understand why you ask for these things, but they are all misconceptions based on misleading marketing. I have lost around four hours writing long replies that disappear, so I will try and keep this one shorter...

 

1) Adding a KX-R level volume control to a DAC would increase the price by over $7,000. Then it would only control the components connected to the DAC. A better place to move the volume control is to the amplifier if you want to reduce the number of boxes.

 

2) You are really asking to combine a preamplifier with a DAC. There are many ways to combine components. I do not think this is the best combination. You can read more about these first two points here:

 

RE: Ping Charles Hansen - Integrated amp question - Charles Hansen - Amp/Preamp Asylum

 

3) True dual mono would mean two power cords, two power switches, two circuit breakers... A lot of expense with almost no return. A better solution is to split the power supplies after the transformer and pre-regulator. Then have separate regulators for each channel. The grounding scheme will make much more difference in channel separation than "dual mono".

 

4) More than 130 dB S/N ratio is only possible if you measure with certain restrictions. The measurement system must be limited from 20 Hz to 20 kHz. Then "A" weighting must be applied. If measured with wideband and no weighting, DSD would barely reach 50 dB S/N ratio. Even the best PCM could only reach 110 dB under those conditions.

 

The number is not so important. The human ear/brain system can easily hear 20 dB or more into the noise floor. I have performed blind tests where we changed the dithering at the 24th LSB level. I was not expecting to hear any difference yet I could correctly identify the dither method 100% of the time! The S/N will be very high - as high as any product ever made. But the actual number will depend more on the measurement method than anything else.

 

4) Rock-solid imaging has nothing to do with channel separation. A top-level LP playback system can have rock-solid imaging. Yet the phono cartridge only reaches a MAXIMUM of 40 to 45 dB in the midband and drops to 20 or 25 dB at the frequency extremes. The power supply design has MUCH more influence on imaging than the measured channel separation numbers.

 

5) LCD screen generate high levels of high-frequency RFI. Bad for audio sound. Cover art is great on an iPod when it is 30 or 40 cm away from your eyes. But to put it on the DAC it may be 10x or 20x further away. To get the same resolution, the screen would need to be 10x larger in each dimension so 100x to 400x as much area!. Then the file size in the metadata would need to be 100x to 400x times larger and the file size would grow significantly. A better solution is a handheld device like the Squeezebox Touch -- except then you are forced to use Wi-Fi to update the data. Wi-Fi puts an RF transmitter in your own house. Now you have increase RFI levels significantly and since it operates at the exact same frequency of a microwave oven, you have turned your house into a low-level microwave oven that is on 24/7/365. Health effects are not yet known, but likely to be bad. We will know in another twenty years...

 

Best regards,

[Paul R]

The z114 was brought back to in-house manufacture, and it shows it too. (grin)

 

It is a lot like the high end audio hardware today - how many "high end" builders actually build their own cases, or wind their own transformers? Spec out their own resistors? Not many I am sure. The microcode in there is all IBM though.

 

On the other hand, the matching DASD for that machine is a DS8800, and it still gripes me to look at the machine. The ethernet switches in the back have a softly glowing logo on them visible through the back door. Only two letters on them instead of three...

 

-Paul

 

 

Wow, how times have changed. I worked for IBM in both Poughkeepsie and Endicott a lifetime ago, when they were primarily a hardware manufacturer. We used to be able to tell how the stock was going to do by walking down to the manufacturing floor and counting the mainframes in final test. The customer's name used to be on a placard above their bay of hardware. These were huge water cooled monsters. Looking inside that server, how much do you think IBM now manufactures, as opposed to bought out? The cabinet?

[Paul R]

Hey Miska - are those computers you design and build designed and built by you alone? You designing the case, power supply, SATA controllers, and all the microcode/firmware? Not bad at all if you are.

 

Or are you just bolting together COTS parts?

 

And actually, yeah- it is a GREAT music playback computer- it streams music very fine to pretty much any number of network players. Mainframes don't break a sweat with any kind of I/O music playback could put on them.

 

Budget and environmental concerns tend to make me choose the second best machine though, and that would be a Mac.

 

-Paul

 

But definitely not the best music playback computer. I build better computers for that purpose.

 

Btw, is that designed by you and you alone? Not bad if it is.

 

I care about what I personally design and develop. Others can do what they want the way they like. As long as I get what I want when I do what I want.

[Miska]

Hey Miska - are those computers you design and build designed and built by you alone? You designing the case, power supply, SATA controllers, and all the microcode/firmware? Not bad at all if you are.

 

Case, power supplies, sometimes the PCBs and partially firmware.

 

Or are you just bolting together COTS parts?

 

To the similar extent Apple does.

 

And actually, yeah- it is a GREAT music playback computer- it streams music very fine to pretty much any number of network players.

 

If you mean by "network player" a UPnP player, then the actual player is software is in the "network player" and the server is just a dumb file server responding to HTTP GET-request. I'm not interested on file servers, I'm only interested on the things that actually deal with audio.

 

Budget and environmental concerns tend to make me choose the second best machine though, and that would be a Mac.

 

Neither one is specifically purpose built for music playback. I have the latest Mac mini, but it's too noisy and not fast enough. (it's fan gets noisy when you run all the DSP stuff on it with constant CPU load around 75%)

 

So I rather have passive cooled Core i7 instead.

[Miska]

By the way, I don't care so much about how the case looks like, I don't need to impress anyone with how it looks, I'm fine with plain simple utilitarian looks:

 

But when necessary I design PCBs and stuff, like this A/D/A board with simultaneous 192/24 PCM + DSD128 (12x12").

(MCU interface board is separate, there are two ribbon cable connectors for it)

 

And no, I don't sell any of this stuff. It's all just for myself. All because things I want are not available on the market.

 

But if you want better computer for music playback than Mac Mini, look for example at

Fit-PC Intense PC Pro

[Paul R]

Teach me to reply to sarcasm with sarcasm....

 

That is a nice PC, but not purpose built for music either.

 

I have yet to hear the fan noise ar all on a 2012 i5 Mac Mini, even running with Convolution. Or running Windows with XXHighEnd or JRMC.

 

the Mac Mini i5 is about half the cost of the PC you linked to configured with Windows. Or when configured with the same i7 processor, 4 gigs of RAM, and a terabyte drive, for $799, same as a barebones version of the PC you linked to withoyt drive, RAM, or OS.

 

So I don't buy your fan noise argument at all. And this us also a great example of why I don't buy the argument that MACs are overpriced either. High quality costs more. Even with PCs.

 

Also drives a $2499 QB9 just fine... Which is also priced the way it is largely because of quality. I expect were Ayre to sell 10 million of them, they would be able to realize an economy if scale like Apple, and the price for a QB9 would be closer to $999. That is purely guessing though.

 

Would probably buy or build an i7 CAPS machine for music, if I wanted to try and improve on the music capability of the Mini, and that would be pretty much because of the SOTA USB card.

 

 

 

Paul

 

 

By the way, I don't care so much about how the case looks like, I don't need to impress anyone with how it looks, I'm fine with plain simple utilitarian looks:

[ATTACH=CONFIG]5994[/ATTACH]

 

But when necessary I design PCBs and stuff, like this A/D/A board with simultaneous 192/24 PCM + DSD128 (12x12").

[ATTACH=CONFIG]5995[/ATTACH]

(MCU interface board is separate, there are two ribbon cable connectors for it)

 

And no, I don't sell any of this stuff. It's all just for myself. All because things I want are not available on the market.

 

But if you want better computer for music playback than Mac Mini, look for example at

Fit-PC Intense PC Pro

[mav52]

 

Neither one is specifically purpose built for music playback. I have the latest Mac mini, but it's too noisy and not fast enough. (it's fan gets noisy when you run all the DSP stuff on it with constant CPU load around 75%)

.

 

Interesting, never heard the fan noise on my latest MacMini

[barrows]

On mini fan noise: I have no reason to doubt Miska's observations, remember, he said he is running complex DSP which has the processor at 75%, I think anyone running the processor in a mini at 75% for any length of time is gong to hear the fan.

 

And on the QB-9, it is very important to understand that Ayre builds all of its products here in Boulder, Colorado, and they also select as many American made components as is reasonable to do. Additionally, they are well known in this community as good company to work for, who treats their employees fairly, with appropriate compensation and benefits. There is no question that they could produce similar products (but perhaps not with the same consistent attention to quality and detail) for less by moving production to China, but they will not do that. I understand that some consumers could care less where a product is built, and only want the cheapest possible price, but personally, I am happy to pay more for things made in this country, to a higher quality standard, by employees who are treated well, and, as such, be contributing to the economy here.

[Claude]

thanks Charles,

Thats what I have been doing the last two years. Forum Talk useally leads to more Audionervosa which does not necessarily lead to better sound.

so i will Listen and wait, rather then wait and listen.

claudius

 

Dear Claude,

 

We have only shipped about 20 of the new units so far. I don't think that most of these people read these forums -- they are too busy listening to their music! :-)

 

Best,

Charlie Hansen

Ayre Acoustics, Inc.

[Paul R]

Eh- we have a philosophical difference I suspect. Playing music on my 2012 i5 Mac Mini runs the processor cores at a negligible percentage of the available CPU cycles. That would be about 2% on one core, and 0% on the other three. That's the lowest cost lowest powered Mac Mini, and it is far more powerful than any other $599 OTS computer I have seen in the Microsoft, BestBuy, or local computer shops.

 

Running on a quad Core i7, getting 75% processor usage on all eight cores , that is one HELL of a lot of DSP, and if you need that much, there is something wrong in the software or audio design.

 

There is no way in the world playing music with any of our players, even upsampling output to 24.192 or converting DSD files to PCM is going to drive even the Mac Mini I have to 75% processor usage.

 

-Paul

 

 

On mini fan noise: I have no reason to doubt Miska's observations, remember, he said he is running complex DSP which has the processor at 75%, I think anyone running the processor in a mini at 75% for any length of time is gong to hear the fan.

 

And on the QB-9, it is very important to understand that Ayre builds all of its products here in Boulder, Colorado, and they also select as many American made components as is reasonable to do. Additionally, they are well known in this community as good company to work for, who treats their employees fairly, with appropriate compensation and benefits. There is no question that they could produce similar products (but perhaps not with the same consistent attention to quality and detail) for less by moving production to China, but they will not do that. I understand that some consumers could care less where a product is built, and only want the cheapest possible price, but personally, I am happy to pay more for things made in this country, to a higher quality standard, by employees who are treated well, and, as such, be contributing to the economy here.

[silvertone]

 

1) Adding a KX-R level volume control to a DAC would increase the price by over $7,000. Then it would only control the components connected to the DAC. A better place to move the volume control is to the amplifier if you want to reduce the number of boxes.

 

2) You are really asking to combine a preamplifier with a DAC. There are many ways to combine components. I do not think this is the best combination. You can read more about these first two points here:

 

RE: Ping Charles Hansen - Integrated amp question - Charles Hansen - Amp/Preamp Asylum

 

Best regards,

 

Charles,

 

Thank you for taking the time to reply.

 

Even considering the $7k price increase that a KX-R volume would add, I still think this would be a more cost effective alternative for us consumers. The other alternative being a KX-Rpreamp plus the DAC. You probably do your homework with the target audience for your products, but I'd think there are plenty customers like myself waiting on the sidelines for a great, no compromises, digital preamp. The analog inputs would be the cherry on top.

 

As far as I know the only DACs in the market with a respectable analog volume control implementation are the MSB DACs, but I'm not willing to cough up $30k + on a DAC. They have a lower priced unit but it's singled ended, no fully balanced. DCS new Vivaldi DAC goes for $30K as well, but again, a digital volume control as well. At that price range, I'd expect more than that.

 

In regards to adding at least one set balanced analog inputs, Classe was able to do this successfully in their CP-800 Preamp with a modest budget. So, it is doable to do this and retain excellent measured performance (see Stereophile's measurements of the CP-800 in their reviews). The only reason I wouldn't buy that unit (it sounds great by the way) is because I'm willing to pay more for a similar product with a more a transparent analog volume control implementation, and higher bit resolution. I think the Classe has only about 19 bits of resolution. I guess the question would be, is it feasible for Ayre to design such a product and sell it for about $15k to $17k MSRP?

[Miska]

That would be about 2% on one core, and 0% on the other three.

 

Bit perfect playback should be way below 1% on all cores. But that's complete waste of CPU time, because rest would go to running OS' idle NOP-loop.

 

Running on a quad Core i7, getting 75% processor usage on all eight cores , that is one HELL of a lot of DSP, and if you need that much, there is something wrong in the software or audio design.

 

There is no way in the world playing music with any of our players, even upsampling output to 24.192 or converting DSD files to PCM is going to drive even the Mac Mini I have to 75% processor usage.

 

Well, you are free to do it better if you can, without making any compromises quality. DAC chips make lot of compromises in their digital filter and modulator design and that's why I don't use those.

 

Especially digital room correction filter processing for DSD can be heavy.

 

And yeah, my algorithms utilize SSE instructions. (FMA3 support is also kind of ready, but it takes a bit time before Haswell, Piledriver and Trinity become popular enough)

[Miska]

DCS new Vivaldi DAC goes for $30K as well, but again, a digital volume control as well. At that price range, I'd expect more than that.

 

What's wrong with digital volume control? It can easily have lower noise than any analog volume control and much better accuracy. I wouldn't want to add any extra analog components on the signal path, those can only degrade quality and add noise.

 

And at least with DSD digital volume control you cannot "lose bits" because there's only one to begin with...

[Miska]

In short: anything less than 100% is just under-utilizing your CPU...

[beetlemania]

Forum Talk useally leads to more Audionervosa which does not necessarily lead to better sound.

 

Indeed! There are many worthwhile tips to be found in these forums but also a great hazard of heightened nervosa. Before getting into computer audio, I fretted over computer choices. But I've found that computer-side tweaks make little to no difference in sonic quality. Maybe the QB-9 is "immune" to such things or maybe my ears aren't so good. Having done my own trials, I'm confident that my system sounds about as good as possible without upgrading my speakers or amp.

[mav52]

thanks Charles,

Thats what I have been doing the last two years. Forum Talk useally leads to more Audionervosa which does not necessarily lead to better sound.

so i will Listen and wait, rather then wait and listen.

claudius

 

Pretty true, just makes you spend more money searching for that sound...

[hifitubes]

Indeed! There are many worthwhile tips to be found in these forums but also a great hazard of heightened nervosa. Before getting into computer audio, I fretted over computer choices. But I've found that computer-side tweaks make little to no difference in sonic quality. Maybe the QB-9 is "immune" to such things or maybe my ears aren't so good. Having done my own trials, I'm confident that my system sounds about as good as possible without upgrading my speakers or amp.

 

Are you kidding? Heightened nervosa here? Have you been to the XXXHighEnd forum, or worse...the JPLAY forum?

 

Someone should start a poll on how many listening rooms have been at least half-properly treated before any type of tweakage ensues.

[Charles Hansen]

Charles,

 

Thank you for taking the time to reply.

 

Even considering the $7k price increase that a KX-R volume would add, I still think this would be a more cost effective alternative for us consumers. The other alternative being a KX-Rpreamp plus the DAC. You probably do your homework with the target audience for your products, but I'd think there are plenty customers like myself waiting on the sidelines for a great, no compromises, digital preamp. The analog inputs would be the cherry on top.

 

As far as I know the only DACs in the market with a respectable analog volume control implementation are the MSB DACs, but I'm not willing to cough up $30k + on a DAC. They have a lower priced unit but it's singled ended, no fully balanced. DCS new Vivaldi DAC goes for $30K as well, but again, a digital volume control as well. At that price range, I'd expect more than that.

 

In regards to adding at least one set balanced analog inputs, Classe was able to do this successfully in their CP-800 Preamp with a modest budget. So, it is doable to do this and retain excellent measured performance (see Stereophile's measurements of the CP-800 in their reviews). The only reason I wouldn't buy that unit (it sounds great by the way) is because I'm willing to pay more for a similar product with a more a transparent analog volume control implementation, and higher bit resolution. I think the Classe has only about 19 bits of resolution. I guess the question would be, is it feasible for Ayre to design such a product and sell it for about $15k to $17k MSRP?

 

Hello Silver,

 

You are welcome. As you have noticed, I am always glad to "share" my opinions. I suppose I like to hear myself talk like some of the other here posters also do... :-)

 

I think that an R level multi-input DAC will look a lot like the KX-R and probably cost between $15,000 and $20,000 without adding an analog volume control. So with an analog volume control it would be between $22,000 and $27,000 -- and that assumes that there is enough room to include it without having to expanding the chassis size. If the chassis size increased, that could increase the retail cost of the product another $3,000 to $5,000.

 

The Classe uses a Cirrus Logic IC-based volume control with an internal high-feedback op-amp. The best thing that one can say about it is that it offers a lot of positions (128) for a low price. As far as performance goes, I would by far rather use a mechanical potentiometer and either forgo the remote operation or just use a cheap Noble motorized pot. They don't sound anywhere near as good as a Shallco switch with top quality metal-film resistors, but their sins are of omission (lack of resolution) rather than comission (grainy and edgy). Classe underpriced their new Delta series product line. When this reality struck them, they were faced with one of two choices -- raise their prices 30% to 40% or move production to China. They chose the latter.

 

You have never addressed why not do the thing that makes the most sense and move the volume control in with the power amplifier? That way one box contains all of the digital circuitry, and the other box (integrated amplifier) holds all of the analog circuitry. Whether you have an all-digital system or some analog sources also, this makes the most sense. The analog box is completely free of high frequency RFI from the high speed clocks. It is just as easy to add six analog inputs as to add two, so there are no limitations if you reconfigure your system or decide to sell it. The digital box could have six or eight inputs also. We have a couple of very nice tricks up our sleeve that we are saving for the R series DAC. I think it it will set a new standard, just as the QB-9 set the standard for (solid-state) computer audio over four years ago (and I think the new updated version will raise the bar again it its price category).

 

Best regards,

[Charles Hansen]

What's wrong with digital volume control?

 

The CTO of ESS (makers of the Saber DAC chips) gives a presentation here:

 

 

explaining the limitations of a digital volume control. The best thing about it is that it is essentially free. The worst thing about it is that even on paper one is limited to no more than between 6 and 40 dB of attenuation before losing resolution (depending on the resolution of the source material). In the real world, sensitive listener will easily hear degradation at perhaps half that much attenuation. A digital volume control will always be a compromise in performance.

 

But in many situations, it is an excellent compromise. However when one wants the absolute highest level of performance, an analog control is the only way to go. Again (just as with PCM versus DSD) implementation is everything. I would much rather listen to your digital volume control, Miska, than I would the Cirrus Logic analog volume control chip, especially if there were a way to adjust the analog the gain of the system (such as the Wadia products used to do) so as to minimize the resolution losses from the digital volume control.

 

Cheers,

[Charles Hansen]

In short: anything less than 100% is just under-utilizing your CPU...

 

You don't want to save a little extra reserve power in case you have a new idea next year that requires more processing power? Then the product becomes obsolete.... I prefer to design products that have a long lifespan in the market place. They can be upgraded when technology improves and also retain their resale value better.

 

Best regards,

[Miska]

You don't want to save a little extra reserve power in case you have a new idea next year that requires more processing power?

 

It is already profiled and optimized to the edge with hand-written SSE assembly, without trading any quality. It can also utilize up to around 24 CPU cores. Of course I add AVX2+FMA3 as CPUs that support it gain market share. That again roughly doubles the operations-per-clock-cycle performance. (and of course I use VFP & NEON on ARM, depending on the particular hardware)

 

But if it doesn't run fast enough for thing X you want, you just need to buy a faster CPU with bigger cache and faster RAM. Some of the things are more limited by cache sizes and RAM speed than the raw CPU speed. Currently I have just dual channel DDR3 1600 MHz RAM myself, but next machine will definitely have DDR3 1866 MHz.

 

So far, I've been pretty well on par on just being able to run things I want on fastest CPUs available on the market.

 

Microsoft Flight Simulator used to be a bit like that, at the time new version was release, there wasn't a fast enough computer to run it with full details and all options on, but such machine was on the market just before release of a new version...

[Miska]

explaining the limitations of a digital volume control. The best thing about it is that it is essentially free. The worst thing about it is that even on paper one is limited to no more than between 6 and 40 dB of attenuation before losing resolution (depending on the resolution of the source material).

 

That's for their implementation. With my oversampled noise-shaped 32-bit PCM output I can get around 233 dB dynamic range for the audio band in digital domain. Internally it is 64-bit floating point which has pretty nice dynamic range from roughly 10^-308 to 10³⁰⁸.

 

For the delta-sigma modulator output at 1024x fs I can do better than that 32-bit PCM.

 

So it would be interesting to hear how do you think someone could hear "loss of resolution". With 24-bit material you can go to -88 dBFS before the noise floor of source material stops dropping down. And with 16-bit source material you can go down to -136 dBFS.

 

Even with dumb flat dither like TPDF and 32-bit resolution all the "loss" in digital domain would be way below thermal noise of 1 ohm resistor. Even with liquid nitrogen cooling in analog volume control you would be still far. And 24-bit source material and 32-bit output using TPDF you can have -48 dBFS before you hit the noise floor of 32-bit.

 

IIRC, with 8 channels in parallel, Sabre output has equivalent noise/resistance of 75 ohm resistor. And with a all discrete delta-sigma DAC you can actually go even lower. Something to watch out for in analog stages following the DAC!

[MikeJazz]

The CTO of ESS (makers of the Saber DAC chips) gives a presentation here:

 

 

explaining the limitations of a digital volume control. The best thing about it is that it is essentially free. The worst thing about it is that even on paper one is limited to no more than between 6 and 40 dB of attenuation before losing resolution (depending on the resolution of the source material). In the real world, sensitive listener will easily hear degradation at perhaps half that much attenuation. A digital volume control will always be a compromise in performance.

 

But in many situations, it is an excellent compromise. However when one wants the absolute highest level of performance, an analog control is the only way to go. Again (just as with PCM versus DSD) implementation is everything. I would much rather listen to your digital volume control, Miska, than I would the Cirrus Logic analog volume control chip, especially if there were a way to adjust the analog the gain of the system (such as the Wadia products used to do) so as to minimize the resolution losses from the digital volume control.

 

Cheers,

 

Very interesting. Thanks for your sharing!

 

The part where he explains that there is some inaccuracy in the digital attenuation...

He starts by talking about a naive/simplistic implementation of a digital volume control..."what would be his consideration if he would talk about a different digital control working at higher resolution". He later answers...

 

Later he talks about that, and seems that good/optimized digital volume controls can be made when they have internal access to the dac, with conclusions at around 10:00.

 

IN this regard, I think it's very logical either to have power dacs, or integrated preamp/dacs with digital volume control...

 

So Charles, you said "However when one wants the absolute highest level of performance".

One knows that the absolute highest level of performance (SOTA level) is very expensive.

It seems that a very well design digital volume control (that can be made at a reasonable cost when integrated into a dac) is only bettered by a very expensive analog control...

If I am not mistaken by the examples you gave, In your view it costs around 7000 USD to build it (which is more that I did spend on the full unit to do dac/control/amplification unit)...

Isn't this a clear case of the law of the diminish returns?

 

***

In my case and experience, I as frustrated some years ago for lack of resolution an low volumes...

My first change was to exchange the pre to a VTL, that was quite an improvement.

But the valved VTL while working very well at low volumes, loosed it's composure when driven very loud, with noticeable distortions...so I looked for a solution to play well in low and in high volumes, at a reasonable cost...

Found it with Lyngdorf.

They state that there are no losses until around 62db (so, 38dB attenuation, I believe) and I never perceived any loss in resolution...

When I lower the volume until around 55 (depending on the source music, of course) I still perceive most of the information, making this a very effective tool for late night listening...

 

Lyngdorf Audio - TDAI 2200 - Power Supply and Volume control

 

But anyway, quite educational...