Compressor with micro-based side chain designing?

[Samuel Groner]

What did you prefer about the MDAC approach with the AD5433  over the old Blackmer VCA circuits?

The signal path of an MDAC is, from an objective point of view, much cleaner (lower noise, distortion) than any analog VCA. I'm not sure which approach is better regarding control signal feedthrough (the MDAC has some glitch charge when switching bits), but it seems to have worked well.

Samuel

 

[JohnRoberts]

What did you prefer about the MDAC approach with the AD5433  over the old Blackmer VCA circuits?

The signal path of an MDAC is, from an objective point of view, much cleaner (lower noise, distortion) than any analog VCA. I'm not sure which approach is better regarding control signal feedthrough (the MDAC has some glitch charge when switching bits), but it seems to have worked well.

Samuel

I've never worked with a MDAC but kicked the tires pretty good on DPOTs. There are two issues with digital gain control modulation. The first affectionately called  "zipper noise"  occurs when gain chainges occur while signal is present so that audio gets modulated by the gain steps (the zipper characterization is from ramping through all the codes versus a single step that could cause a louder click). This gain modulation can be managed by coordinating a gain change with a signal zero crossing, so the gain change gets multiplied times 0V, so  N x 0 = 0. 

A second mechanism that some early DPOTs were guilty of was glitches related to major code transitions. The difference between  0x1000 and 0x0111 is only one LSB but the way the gain control is implemented inside the part causes a current glitch whenever a major bit transition occurs. This is much improved in modern parts, and some premium applications build in the zero cross detection circuitry  (like some mic preamps with digital gain control).

JR

 

[mr coffee]

Thanks much for all the helpful posts - I am still busy digesting all the good information.

Andy,
I went and looked at the Silicon Labs offerings, also  Atmel (AVR,  Cortex0, ARM stuff ). Whew! Sophisticated stuff out there at surprisingly cheap prices. Most looked like overkill for what I'm trying to do, at least to my (admittedly) naive flyby of the datasheets.  500+ pages on a lot of them.  Whew! Analog, take me home!!  ;D

Ethan,
Thanks for the pointer to the 24F family and ideas. I can live with 14 pins, especially if I can do a DIP package and spare myself the horrors of on-board "engineering changes" on SMT boards during circuit development.

And do I read between the lines - both you and John - that you feel OK with the cheap-to-free development tools coming from Microchip these days - the IDE, the on-chip debugging resources, software development support tools, etc.? Has Microchip got their act together better in that department these days? I didn't see a freebie high-level programming package, (maybe I missed it?), but I can stand using  assembler if necessary. Seems I imagine I can tell more about what the code is telling the hardware to do anyway with assembler.  ;D

Sam,
Thanks very much for your reply about your experience using the AD5433 MDAC  for your variable gain element. So it worked quite well in your design - did you have to use a pair of '5433s to balance out the switching charge injection, dc offset changes with gain, and switch r\2r non-linearities ?  Did you need to switch gain with zero-crossings to keep artifacts down?

John,
About the same questions as to Sam about using the DPOTS for a variable gain element, if you were experimenting with using them for more than gradual gain changes\level setting.

Thanks to everyone for being so helpful. I really appreciate your help.

 

[JohnRoberts]

And do I read between the lines - both you and John - that you feel OK with the cheap-to-free development tools coming from Microchip these days - the IDE, the on-chip debugging resources, software development support tools, etc.? Has Microchip got their act together better in that department these days? I didn't see a freebie high-level programming package, (maybe I missed it?), but I can stand using  assembler if necessary. Seems I imagine I can tell more about what the code is telling the hardware to do anyway with assembler.  ;D

Sam,

I am OK enough with microchip to not start over from scratch and learn another brand's instruction set and quirks, but  I am not ready to endorse them to anyone else...

Like I've said, I'm happy with the hardware, not so much the support.

John,
About the same questions as to Sam about using the DPOTS for a variable gain element, if you were experimenting with using them for more than gradual gain changes\level setting.

Thanks to everyone for being so helpful. I really appreciate your help.

I was developing an automatic mixer using one micro to crunch gain. For my first prototype I laid out both THAT VCAs and DPOTs for the gain control. I brought up the DPOTs first and they sounded so good I didn't bother with the VCA,s.

I anticipated needing to coordinate gain changes at least for the loudest channels (In an automatic mixer the loudest channels get the most gain, less loud channels get less gain, etc).  I rolled my own version of SPI so I could push gain to pairs of DPOTs, with the ability to control with the micro when that gain change latched into the DPOT. My plan was to co-ordinate gain changes with zero crossing of the loudest channels, and updating the quieter channels as available. Not only did the DPOTs work better than I expected but I didn't need to perform any fancy tricks with the gain changes just updating them all sequentially at a decent pace.  The gain was updating fast enough, and the gain changes per update were small enough that I did not hear any audible artifacts. I already had zero crossing info from the A/D conversions with enough processor headroom to add zero crossing synchronization if needed, but I didn't hear any need.

This project got back burnered before it was finished so I never got to finalize it, but I was doing all the dog and pony tricks I asked it too...  That one processor was scanning all the front panel switches and controls, provided parallel rear panel I/O and even a multiple unit link capability.  I still have the proto board but haven't turned it on for a couple years...

Even back then I was arguing how cheap it would be to drop in a few codecs and do the whole thing in the digital domain, but now these automatic mixers are becoming free software options inside digital mixers.

JR

 

[mr coffee]

John,
I've been  wading through various DPOT ICs today, and it looks like the trend is toward cheaper, lower voltage designs (like for a cellphone or iPod-type app).  Do you remember which dpots impressed you back when you were messing with them?

Looks to me like the more pro audio quality designs were done a decade ago and are still pretty pricey... $10+ in unit quantity, $20 for the really good ones like the Burr-Brown PGA2310, 2311, 2320 series that do +\- 15 volt supplies and normal audio levels (at least as I conceive of them). The CS3310 is capable of +\-3.5v audio in at really low distortion using +\-5 supplies, for around $10, and they carry on like they have low feedthrough from the gain control port on the datasheet. Several other offerings that could at least do +\-5v supplies and an audio input range that doesn't overload over next to nothing, are now shoved over to the "OBSOLETE" column.  :

I DID find an interesting MAXIM device - DS1882, that looks like it could do a little better audio input range than most of the newer offerings (tolerate +5 to -7 volts of audio in, low distortion at much lower input levels I expect), but it was only 32 steps and looked targeted at personal "consumer electronics". At least it was relatively cheap - a couple of bucks apiece. I think I'll keep it in mind for something less demanding down the line. Pretty cool part - I overlooked it at first because it does non-volatile gain-setting storage, which I wasn't interested in for this app, but turns out there is no comparable part without the non-volatile storage setup and the wider input voltage range. Go figure, right? ???

Maybe Sam will drop  by at some point and share what he had to do to make his AD5543  DAC play nice in a dynamic, gain-following-an-envelope mode, like de-essing.  I'm thinking the parallel interface was advantageous in his design because it cut the latency, (and maybe it made it easier if he had to do a balanced design, too.).

I'm back to thinking a VCA may be the best choice for an effect-type compressor with a lot of gain range and change. Or maybe even a couple THAT VCAs in series to keep them operating in the more desirable parts of their attenuation\gain curve where the signal-to-noise performance is still pretty decent. 

Or anybody else find something in the digital gain control camp that looks interesting that I missed?

Thanks to all.

 

[JohnRoberts]

John,
I've been  wading through various DPOT ICs today, and it looks like the trend is toward cheaper, lower voltage designs (like for a cellphone or iPod-type app).  Do you remember which dpots impressed you back when you were messing with them?

Design is a constant tradeoff between cost-benefit... the parts we were looking at were AD5262. 2x 8 bit DPOTs that would support 15V rails.  30V rails and more bits were pretty expensive. 

Looks to me like the more pro audio quality designs were done a decade ago and are still pretty pricey... $10+ in unit quantity, $20 for the really good ones like the Burr-Brown PGA2310, 2311, 2320 series that do +\- 15 volt supplies and normal audio levels (at least as I conceive of them). The CS3310 is capable of +\-3.5v audio in at really low distortion using +\-5 supplies, for around $10, and they carry on like they have low feedthrough from the gain control port on the datasheet. Several other offerings that could at least do +\-5v supplies and an audio input range that doesn't overload over next to nothing, are now shoved over to the "OBSOLETE" column.  :

I DID find an interesting MAXIM device - DS1882, that looks like it could do a little better audio input range than most of the newer offerings (tolerate +5 to -7 volts of audio in, low distortion at much lower input levels I expect), but it was only 32 steps and looked targeted at personal "consumer electronics". At least it was relatively cheap - a couple of bucks apiece. I think I'll keep it in mind for something less demanding down the line. Pretty cool part - I overlooked it at first because it does non-volatile gain-setting storage, which I wasn't interested in for this app, but turns out there is no comparable part without the non-volatile storage setup and the wider input voltage range. Go figure, right? ???

Maybe Sam will drop  by at some point and share what he had to do to make his AD5543  DAC play nice in a dynamic, gain-following-an-envelope mode, like de-essing.  I'm thinking the parallel interface was advantageous in his design because it cut the latency, (and maybe it made it easier if he had to do a balanced design, too.).

I'm back to thinking a VCA may be the best choice for an effect-type compressor with a lot of gain range and change. Or maybe even a couple THAT VCAs in series to keep them operating in the more desirable parts of their attenuation\gain curve where the signal-to-noise performance is still pretty decent. 

Or anybody else find something in the digital gain control camp that looks interesting that I missed?

Thanks to all.

For the record I am not advocating  DPOT for a general purpose compressor,,, I used it for an automatic mixer that was cut only, with resolution most significant up around unity gain, so a linear 8 bit scale was fine.

If you are going to mess with digital, why not use a codec and do it completely digital...?

JR

 

[joaquins]

...For the record I am not advocating  DPOT for a general purpose compressor,,, I used it for an automatic mixer that was cut only, with resolution most significant up around unity gain, so a linear 8 bit scale was fine.

If you are going to mess with digital, why not use a codec and do it completely digital...?

JR

Because for us young boys analog path is still a nice thing!

Are there any log scaled DPOT or are all linear? of course the loading trick could be used to get a proper definition in the low side but still 8 bit would just one bit more than MIDI!!! of course in a compresor you only look for so much compression you could already know, maybe 20dB is fine and split 20dB into evenish 8 bits and you are good to go. For FF still you need to know but math are easy to know beforehand the attenuation. I guess now there should be some good enough with a few more bits which makes everything smoother in design and final result.

JS

 

[JohnRoberts]

Because for us young boys analog path is still a nice thing!

I'm an old analog dog but an engineer first and when digital can do it cheaper and better it's hard to ignore.

Are there any log scaled DPOT or are all linear? of course the loading trick could be used to get a proper definition in the low side but still 8 bit would just one bit more than MIDI!!!

There have been dedicated volume controls, and the switched resistor gain sets used for digital mic preamps are dB/log scale.

For my application I was operating near the top of the scale , close to unity so the linear resolution was fine... For a compressor it is also doing most of it's work in the top several dB, only for digital faders do you want or need resolution down below -40 dB. 

of course in a compresor you only look for so much compression you could already know, maybe 20dB is fine and split 20dB into evenish 8 bits and you are good to go. For FF still you need to know but math are easy to know beforehand the attenuation. I guess now there should be some good enough with a few more bits which makes everything smoother in design and final result.

JS

It has been years since I looked at this but my speculation is that DPOTs and VCAs are not getting any cheaper, while codecs and  DSP surely are.... If I was a young man... I'd be messing with that. Of course marketing is another issue... Once you start working in the digital domain you get compared to SOTA  conversions. A $1 A/D  chip is probably already equal to or better than the performance of a modern VCA, but the customers will see a $1 A/D and puke...

Life is too short to argue with customers about what is good... (I still have some gold plated RCA jacks, maybe that would help?)

JR 

 

[joaquins]

I'm an old analog dog but an engineer first and when digital can do it cheaper and better it's hard to ignore.

I know, I also like to play around with those, but there is certain charm designing analog, plus I really don't like much programing.
Customers also pay more for studios equipped with tons of analog gear, so no wonder the owner of the studio prefers analog, then we prefer analog too, even more expensive and better in the paper, for an analog pice of s***t there is someone who pays more.

There have been dedicated volume controls, and the switched resistor gain sets used for digital mic preamps are dB/log scale.

For my application I was operating near the top of the scale , close to unity so the linear resolution was fine... For a compressor it is also doing most of it's work in the top several dB, only for digital faders do you want or need resolution down below -40 dB. 

I was looking for >60dB fader at work, we considered every single possibility, VCA and MDACs between may others, every single one has it own trades off of course and at the end of the day price wasn't changing too bad IIRC, maybe the DPOT may be more expensive but needs nothing around, the VCA will take a DAC plus quite a few added external components.

It has been years since I looked at this but my speculation is that DPOTs and VCAs are not getting any cheaper, while codecs and  DSP surely are.... If I was a young man... I'd be messing with that. Of course marketing is another issue... Once you start working in the digital domain you get compared to SOTA  conversions. A $1 A/D  chip is probably already equal to or better than the performance of a modern VCA, but the customers will see a $1 A/D and puke...

I know, as I said before, there is people paying more for analog and I enjoy more designing analog, so, not much to think about. On the other hand, who would buy a digital compressor which needs wires has converters and lot of things instead of using a plugin, live or studio, this days? Maybe designing plugins is the way to go but then we are out of job as EE, maybe better became  a programmer, but I will really miss my iron and wire crimpers!

Life is too short to argue with customers about what is good... (I still have some gold plated RCA jacks, maybe that would help?)

JR

LOL about gold plated RCA... I'll take that!

JS

 

[JohnRoberts]

I'm an old analog dog but an engineer first and when digital can do it cheaper and better it's hard to ignore.

I know, I also like to play around with those, but there is certain charm designing analog, plus I really don't like much programing.
Customers also pay more for studios equipped with tons of analog gear, so no wonder the owner of the studio prefers analog, then we prefer analog too, even more expensive and better in the paper, for an analog pice of s***t there is someone who pays more.

There is plenty of low cost analog gear already. That market is not my idea of a growth opportunity.

There have been dedicated volume controls, and the switched resistor gain sets used for digital mic preamps are dB/log scale.

For my application I was operating near the top of the scale , close to unity so the linear resolution was fine... For a compressor it is also doing most of it's work in the top several dB, only for digital faders do you want or need resolution down below -40 dB. 

I was looking for >60dB fader at work, we considered every single possibility, VCA and MDACs between may others, every single one has it own trades off of course and at the end of the day price wasn't changing too bad IIRC, maybe the DPOT may be more expensive but needs nothing around, the VCA will take a DAC plus quite a few added external components.

For a -60dB fader unless you get a dedicated volume control taper the linear law will be unacceptable at the bottom. VCAs have the nice characteristic of turning down their own noise floor along with the input signal.  Yes the VCA requires more glue to control, DPOT just needs SPI or I2S.

It has been years since I looked at this but my speculation is that DPOTs and VCAs are not getting any cheaper, while codecs and  DSP surely are.... If I was a young man... I'd be messing with that. Of course marketing is another issue... Once you start working in the digital domain you get compared to SOTA  conversions. A $1 A/D  chip is probably already equal to or better than the performance of a modern VCA, but the customers will see a $1 A/D and puke...

I know, as I said before, there is people paying more for analog and I enjoy more designing analog, so, not much to think about. On the other hand, who would buy a digital compressor which needs wires has converters and lot of things instead of using a plugin, live or studio, this days? Maybe designing plugins is the way to go but then we are out of job as EE, maybe better became  a programmer, but I will really miss my iron and wire crimpers!

Sorry if you don't like the trend . I begged some friends who were making high quality analog products to get involved with digital technology and they resisted. I worked with them on a digitally controlled analog product, but it was too little too late, as their main market was eroded by lower cost high performance digital. 

There will probably be an analog niche in recording for some time. Some young people have started buying vinyl, while I see that as a dead cat bounce and not a mass market revival. Vinyl may persist as a niche. like tube amps today.

Life is too short to argue with customers about what is good... (I still have some gold plated RCA jacks, maybe that would help?)

JR

LOL about gold plated RCA... I'll take that!

JS

Last I heard the gold plated jacks were not selling well.

JR

 

[PRR]

Off the intended topic, but I ran across this plan and it begs for revival.

Wireless World March 1970. Grossi and Marcus were copying tapes and needed to find peak levels. Manual metering and potting did not satisfy them. This scheme, they roll the source, let the box find the gain which leaves the peak level at nominal 1V. Then they rewind the source and do the copy at the fixed gain.

The 6-bit 1dB steps digitally controlled amplifier is a marvel of 1970's thinking. There's some stage-swopping to get good overload.

The transistors on the top work exactly the same as the ones on the bottom; drawn for a pretty layout instead of gestalt understanding. Does not help there is an error on one transistor.

They state "4000dB per second", which sounds fast, but is really 4dB per mS or 5ms for 20dB. Plenty fast for many uses, and I'd think it could be far faster if desired (reduce C9 for starts).

 

[PRR]

The peak detector and overload counter is nearly as simple as could be. The 3rd transistor is not a JFET but a Unijunction-- ask your grandfather.

And this was the cusp of when you would do the counting in TTL/CMOS. Evidently their parts-bins had years of 2N3708 but they had not yet ordered many chips.

 

[joaquins]

That looks like an MDAC, with parallel instead of ISP, but still! Of course here we are trying to get much better resolution than 1dB/step but it could be rescaled or added stages to get where you want.

I didn't know about those UJT, looks like a JFET but with the gate builded on the channel instead the other way around... The negative resistance zone makes it attractive to play with, I wonder if I can find some locally... 

Thanks for sharing PRR!

JS

 

[mr coffee]

John,

For the record I am not advocating  DPOT for a general purpose compressor

Never thought you meant that for a minute. But your remarks and Sam's got me thinking more about what I wasn't considering. 

DPOTs do look like they could do some cool stuff in controlling programmable analog circuitry - something I might want to put together down the road, though. Thanks for getting me to check back in on what has been happening in that area.

VCAs have the nice characteristic of turning down their own noise floor along with the input signal. 

That's why I think the codec\all digital approach isn't a match for an analog VCA controlled by a nice sidechain ... when the range of gain control is really wide, like when compression is being used as an effect, which is mostly what I'm aiming at here - and  had hoped someone else around here might be interested in, although I've pretty well figured out by now I'm off on my own trip, unfortunately. <sniff> Nothing new, there. <grin>

BTW - I read that the RNC uses an 8-bit processor for it's sidechain - i guess with analog smoothing on the CV coming out of the micro.

PRR -

but I ran across this plan

Offtopic-schmoftopic.  Hey, I still remember sitting in the library reading that thing.
What a trip! Google should map your brain and plug you and your vast library of schematics and papers into it's database.  8)

Evidently their parts-bins had years of 2N3708 but they had not yet ordered many chips.

Hilariously put, but sooo true. 

Probably was just the ticket for copying tracks off LPs onto cassettes. Hey, Dolby B ruled back then, right?

Aww-oh...not another flashback... I think I need to go lay down now ;D ;D ;D

 

[joaquins]

I'd would be in but I have no time right now... this is an interesting project for me.

If you want to get a wider output range together with lower noise is hard to archive better than a 24 bit DAC, since the limiting factor is usually the analog circuit after it. You could get even better with some DACs which allow you to change the voltage reference in the field, so if you get 20dB attenuation you could change the reference voltage to 1/10th and still use all the definition, or just use it as a MDAC for the slow compression (signal converted, side chain analog voltage as reference) and make the fast limiting in digital where you have better control of the response and can get lower distortion and still not miss a peak.

JS

 

[JohnRoberts]

VCAs have the nice characteristic of turning down their own noise floor along with the input signal. 

That's why I think the codec\all digital approach isn't a match for an analog VCA controlled by a nice sidechain ... when the range of gain control is really wide, like when compression is being used as an effect, which is mostly what I'm aiming at here - and  had hoped someone else around here might be interested in, although I've pretty well figured out by now I'm off on my own trip, unfortunately. <sniff> Nothing new, there. <grin>

BTW - I read that the RNC uses an 8-bit processor for it's sidechain - i guess with analog smoothing on the CV coming out of the micro.

I did my first tuner using an 8 bit micro, and never again... the world comes in more than 256 colors or frequencies or whatever... but for driving a VCA keep in mind you are operating a log control voltage. So for a compressor that might be running 30-40 dB of total gain range, 8-bit DACs give you  0.15 dB step size,,,

PRR -

but I ran across this plan

Offtopic-schmoftopic.  Hey, I still remember sitting in the library reading that thing.
What a trip! Google should map your brain and plug you and your vast library of schematics and papers into it's database.  8)

Evidently their parts-bins had years of 2N3708 but they had not yet ordered many chips.

Hilariously put, but sooo true. 

Probably was just the ticket for copying tracks off LPs onto cassettes. Hey, Dolby B ruled back then, right?

Aww-oh...not another flashback... I think I need to go lay down now ;D ;D ;D

I had a job interview back in 1970 when I was shown a two transistor flip flop and had to describe how it worked. I got that job but was drafted before I could take it. 

That limiter circuit is typical of the time, but today a cheap micro could do most of that with a couple dollar IC.

JR

 

[Andy Peters]

BTW - I read that the RNC uses an 8-bit processor for it's (sic) sidechain

I believe it's one or another variant of the 8051, which is indeed an 8-bit processor. (Nobody uses the original Intel 8051 any more.)

It's a common misconception that an 8-bit processor is limited to doing only 8-bit math.  That is totally false. You can do math with any arbitrary operand/result precision. It just takes more time to process operands larger than the native word size. A 32-bit processor can do a 32-bit add in one clock cycle. An 8-bit processor will take maybe a dozen cycles to do the same operation. The end result will be exactly the same.

As long as you can do the calculation in the required time, the processor's native word length doesn't matter.

And if you write your code in C, then the compiler handles the horrible details of register-swapping and memory access for you. All you do is write foo = bar + bletch and you call it done. I don't do anything in assembly code. There is no point because the compilers are both Really Good and Free.

-a

 

[kambo]

As long as you can do the calculation in the required time, the processor's native word length doesn't matter.

1+

 

[JohnRoberts]

I paid a few hundred dollars for my C compiler when I started my first 8 bit design and never used it. For what I was trying to accomplish writing in assembler was just more cause and effect (perhaps an old analog affectation)... I ran out of bandwidth soon enough (that first platform ran from a 4 MHz crystal, and that gets divided down for the internal clock).

I like the idea of writing in C and letting someone else worry about the details but I have enough trouble with the details when I directly push bits back and forth. I can't imagine troubleshooting the details with another level of abstraction in the middle.

Of course if I was managing a group of programmers I'd want them to be time efficient, luckily I'm not.

Me personally I remember with pain all the extra steps of handling more than 8 bit data with an 8 bit wide path. Lots of opportunities for mistakes too.

Nowadays even 32 bit silicon is pretty cheap. While I can paint my world adequately with 65K colors.

JR

 

[PRR]

> looks like an MDAC

It is; but unlike most MDACs it is *dB* input (yes, parallel).

> That limiter circuit is typical of the time

The circuit details are so 1970, but the overall conception seems very novel to me.

Moog/ARP were doing dB-scale VCAs but analog control.

1dB steps could be brutal for a live real-time controller. 1/2 and 1/4 dB steps could be added without more gain-stages, same as TR12 and TR13 do small-steps in one node. This also gets you to 8 bits, which suits many microcomputer sizes. If you must use a 4-pin PIC, then add a shift-register to offload your 8 bits through one pin.

As a real-time limiter, you let the gain-reduce work about like the original, selecting the UJT buzz-rate to give a decent attack rate. That will only reduce, never release. For some work I have done, this could be acceptable. Generally you will want to feed slow-gain-up pulses so gain sneaks back up. Then you also want a max gain-up (unless you want your noise-level to rise, as in speech-only AGC). And then you think about faster release for isolated transients and slower release if the signal has slammed a lot recently. Also a gain-hold function for when the signal drops into the noise (between musical movements, or pauses in a movie's sparse soundtrack). All we do with voltage processing can be done just as well with pulses. Perhaps better, because analog voltage is hard to store for many seconds, while the flops will just sit there.

> copying tracks off LPs onto cassettes. Hey, Dolby B ruled back then, right?

You must be a child of the 1970s.... you do not remember them.

Philips Cassette existed in 1970 but had not found its niches and was rarely seen. Dolby was still costly studio gear, Dolby B came even later.

I am thinking "language lab". The tapes wear out. You make working copies and keep the masters in the cupboard. The masters arrive with all different recorded levels. The lab users whine about having to use their volume control knobs (some cheap labs didn't give the users knobs). Given labor (student workers), the usual approach was to re-record all tapes at MAX clean level. Given lazy or slovenly labor, a peak-finder is a great help.

I did similar work but with music-appreciation tapes. These were usually fairly decent levels, the playbacks did have knobs, and it was one more add-on job-duty for me so I didn't really give a drat.