Another new signal processing book

On 08/06/2014 11:50 AM, robert bristow-johnson wrote:
> On 8/5/14 10:50 PM, Steve Underwood wrote:
>> On 08/06/2014 10:18 AM, Les Cargill wrote:
>>> A lot of early digital machinery was badly done. Oversampling converters
>>> came later. One of the '90s DAT machines had legendary bad converters.
>>
>>
>> I never found an early CD player that didn't sound truly awful. I tried
>> a number of early ones in an AB comparison with vinyl, and they all
>> sounded far worse than the vinyl - although the lack of surface noise
>> was a wonderful thing :-). I don't think consumer CD players started to
>> do what they should until the first generation with sigma-delta
>> converters.
>
> it wasn't just the early CD players, but also the early CDs.  some of
> the early recordings were with 14-bit ADCs.  i remember in 1989 Stan
> Lipshitz found a CD recording of a symphony that had a bad bit higher
> than the LSB.  you could hear it toggle during flute solos or other
> quiet moments in the song.  maybe it was a dead zone.  quite nasty.
>
> anyway, with 16-bit and 44.1 kHz carved into stone (sorta prematurely),
> it was both amazing and fortunate that, with the use of dither, they
> were able to squeeze more dynamic range outa CDs without having to
> change the format and obsolete the earlier pressed CDs.  there is even a
> way to recover the 4.77 dB lost with additive dither by subtracting it
> out and playback through a more-than-16 bit DAC.
>
>> Thankfully studio equipment had been using much better
>> converters since way before the first CD players appeared.
>
> i dunno about that.  16-bit conventional (like from Analogic) were what
> i was seeing in 1982.

In the late 70s some people did get a good honest 16 bits from their 
audio speed ADCs, especially if they were ovenized.

>> A lot of early digital recordings are very good.
>
> and some were very bad.

Of course. A lot of pretty much anything is very badly done.

>> As far as I know all the DAT machines used sigma-delta converters for
>> both ADC and DAC.
>
> the *early* studio DATs?  really?
>
>> However, some had bizarrely low oversampling ratios.
>> There was something really crooked about the spec sheets for those
>> machines.
>
> there is still a lot of crooked specsmanship goin on.

Yep. Like Dyson saying their brushless DC motor vacuum cleaners have 
zero carbon emissions.

Regards,
Steve

On 8/6/14 12:28 AM, Steve Underwood wrote:
> On 08/06/2014 11:50 AM, robert bristow-johnson wrote:
>> On 8/5/14 10:50 PM, Steve Underwood wrote:
>>> On 08/06/2014 10:18 AM, Les Cargill wrote:
>>>> A lot of early digital machinery was badly done. Oversampling
>>>> converters
>>>> came later. One of the '90s DAT machines had legendary bad converters.
>>>
>>>
>>> I never found an early CD player that didn't sound truly awful. I tried
>>> a number of early ones in an AB comparison with vinyl, and they all
>>> sounded far worse than the vinyl - although the lack of surface noise
>>> was a wonderful thing :-). I don't think consumer CD players started to
>>> do what they should until the first generation with sigma-delta
>>> converters.
>>
>> it wasn't just the early CD players, but also the early CDs. some of
>> the early recordings were with 14-bit ADCs. i remember in 1989 Stan
>> Lipshitz found a CD recording of a symphony that had a bad bit higher
>> than the LSB. you could hear it toggle during flute solos or other
>> quiet moments in the song. maybe it was a dead zone. quite nasty.
>>
>> anyway, with 16-bit and 44.1 kHz carved into stone (sorta prematurely),
>> it was both amazing and fortunate that, with the use of dither, they
>> were able to squeeze more dynamic range outa CDs without having to
>> change the format and obsolete the earlier pressed CDs. there is even a
>> way to recover the 4.77 dB lost with additive dither by subtracting it
>> out and playback through a more-than-16 bit DAC.
>>
>>> Thankfully studio equipment had been using much better
>>> converters since way before the first CD players appeared.
>>
>> i dunno about that. 16-bit conventional (like from Analogic) were what
>> i was seeing in 1982.
>
> In the late 70s some people did get a good honest 16 bits from their
> audio speed ADCs, especially if they were ovenized.
>

they couldn't have been cheap.  in the late 70s, i was still dealing 
with 12 and 14-bit DACs and i made a (very slow, not audio speed) ADC 
with such a DAC, a good comparator, and a little 
successive-approximation code in the Mot 6809.  those DACs were still 
about $50 each which could buy 4 or 5 dinners for two, not just one.

>>> A lot of early digital recordings are very good.
>>
>> and some were very bad.
>
> Of course. A lot of pretty much anything is very badly done.
>
>>> As far as I know all the DAT machines used sigma-delta converters for
>>> both ADC and DAC.
>>
>> the *early* studio DATs? really?
>>
>>> However, some had bizarrely low oversampling ratios.
>>> There was something really crooked about the spec sheets for those
>>> machines.
>>
>> there is still a lot of crooked specsmanship goin on.
>
> Yep. Like Dyson saying their brushless DC motor vacuum cleaners have
> zero carbon emissions.

well, i guess he coulda built the vacuum cleaner with a Briggs&Stratton 
motor instead.


-- 

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

On 08/06/2014 12:59 PM, robert bristow-johnson wrote:
> On 8/6/14 12:28 AM, Steve Underwood wrote:
>> On 08/06/2014 11:50 AM, robert bristow-johnson wrote:
>>> On 8/5/14 10:50 PM, Steve Underwood wrote:
>>> i dunno about that. 16-bit conventional (like from Analogic) were what
>>> i was seeing in 1982.
>>
>> In the late 70s some people did get a good honest 16 bits from their
>> audio speed ADCs, especially if they were ovenized.
>>
>
> they couldn't have been cheap.  in the late 70s, i was still dealing
> with 12 and 14-bit DACs and i made a (very slow, not audio speed) ADC
> with such a DAC, a good comparator, and a little
> successive-approximation code in the Mot 6809.  those DACs were still
> about $50 each which could buy 4 or 5 dinners for two, not just one.

Even then $50 wasn't a lot for a professional application, even if it 
was a week's pay for many people.

Why would you use a 6809, rather than a hardware SAR register, like the 
ones AMD made? Burr Brown made DACs that would turn into a respectable 
audio speed ADC with the addition of one of those SAR registers. 
However, the converters I know that gave 16 bits or more were laser 
trimmed at one temperature and ovenized. They were not $50. :-)

Regards,
Steve

On Wed, 06 Aug 2014 10:53:08 +0800, Steve Underwood <steveu@dis.org>
wrote:

   [Snipped by Lyons]
>
>Strange title. It gives no indication that its not for DSP engineers. I 
>guess going the other way and calling it DSP for Dummies would have been 
>worse. :-\
>
>Steve

Hi Steve,
   I had no control over the book's title.

You're right, "The Essential Guide to ABC" 
book's are Prentice Hall's version of the 
"ABC for Dummies" books.

[-Rick-]
 

On Tue, 05 Aug 2014 21:34:03 -0500, Tim Wescott
<tim@seemywebsite.really> wrote:

   [Snipped by Lyons]

Hi Tim,

>Perhaps a really happy engineer who got help from his wife in Marketing?

What are the chances of that?

>I dunno.
>
>Kim's a solid guy, though, so you may not want to grade him DOWN for the 
>review.

I suspect you're right.
I doubt that Fowler would refer to himself as "an 
accomplished and richly experienced lab technician."

Fowler was probably happy to read that review and 
thought no more about it.

[-Rick-] 

On 8/6/14 1:29 AM, Steve Underwood wrote:
> On 08/06/2014 12:59 PM, robert bristow-johnson wrote:
>> On 8/6/14 12:28 AM, Steve Underwood wrote:
>>> On 08/06/2014 11:50 AM, robert bristow-johnson wrote:
>>>> On 8/5/14 10:50 PM, Steve Underwood wrote:
>>>> i dunno about that. 16-bit conventional (like from Analogic) were what
>>>> i was seeing in 1982.
>>>
>>> In the late 70s some people did get a good honest 16 bits from their
>>> audio speed ADCs, especially if they were ovenized.
>>>
>>
>> they couldn't have been cheap. in the late 70s, i was still dealing
>> with 12 and 14-bit DACs and i made a (very slow, not audio speed) ADC
>> with such a DAC, a good comparator, and a little
>> successive-approximation code in the Mot 6809. those DACs were still
>> about $50 each which could buy 4 or 5 dinners for two, not just one.
>
> Even then $50 wasn't a lot for a professional application, even if it
> was a week's pay for many people.

well, i was an undergrad and graduate student at the time.  i think my 
stipend at the time exceeded $50/week, but not much more than $100.  i 
think i was living on $450/month and waived tuition.

> Why would you use a 6809, rather than a hardware SAR register, like the
> ones AMD made?

well, quoting our sage Jerry: "Engineering is the art of making what you 
want from things you can get."

we already had these microprocessor boards with 6802 and others with 
6809.  i was doing a project for a gerontologist where he was using a 
pressure transducer to measure some property of blood serum.  it was 
sub-audio signal processing and my sample rate was something like 10 Hz 
so the 8-bit fixed-point processing of a cheap micro was sufficient to 
do the algorithm in 0.1 sec with time to spare.  (lotsa add-with-carry 
instructions.)  that time-to-spare was used for the successive 
approximation.  the only extra hardware i needed was a comparator to 
feed back to the MPU.  in fact, we used a S/H so that the same DAC could 
be used for *both* input and output.

the output would go to a chart recorder of the time.  the algorithm was 
predicting the equilibrium pressure of this process modeled as a very 
slow first-order process.

> Burr Brown made DACs that would turn into a respectable
> audio speed ADC with the addition of one of those SAR registers.
> However, the converters I know that gave 16 bits or more were laser
> trimmed at one temperature and ovenized. They were not $50. :-)

they were much more expensive, right?


-- 

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

Rick Lyons <R.Lyons@_BOGUS_ieee.org> wrote in 
news:aer2u99e1ircjmb35esdukag7muhedjasn@4ax.com:

> 
> 
> Hey Tim,
>   I took a look at the Amazon page for Kim Fowler's 
> "Electronic Instrument Design: Architecting for 
> the Life Cycle" book.  It looks like an interesting 
> book.  Not interesting enough to spend $145, but 
> interesting nonetheless.
> 
> Then I read that book's single Customer Review.
> Tim, I'm willing to bet a six-pack of Pilsner 
> Urquell that the single Customer Review is 
> fraudulent.  That is, I believe the review was 
> posted by someone who is somehow connected with 
> the publication of that book.
> 
> The review is beautifully well-written, too beautiful.
> What reader would write that the book contains, 
> "...the wisdom of an accomplished and richly 
> experienced lab technician"?
> 
> Engineers don't write like that, but marketing 
> people do.  Yep, I think I'm making a safe bet.
> 
> I was badly burned once by believing the Amazon 
> reviews of a signal processing book.  So now 
> I read those reviews *VERY* carefully!
> 
> [-Rick-]

You may be right.

Your insight is supported by what I call the Lyon's exception:

"All engineers can't write or can barely write"

I strive to be the latter. 

Al 

On Wed, 06 Aug 2014 10:57:54 +0800, Steve Underwood wrote:

> On 08/06/2014 08:11 AM, Rick Lyons wrote:
>>
>>
>> Hey Tim,
>>    I took a look at the Amazon page for Kim Fowler's
>> "Electronic Instrument Design: Architecting for the Life Cycle" book. 
>> It looks like an interesting book.  Not interesting enough to spend
>> $145, but interesting nonetheless.
>>
>> Then I read that book's single Customer Review.
>> Tim, I'm willing to bet a six-pack of Pilsner Urquell that the single
>> Customer Review is fraudulent.  That is, I believe the review was
>> posted by someone who is somehow connected with the publication of that
>> book.
>>
>> The review is beautifully well-written, too beautiful.
>> What reader would write that the book contains, "...the wisdom of an
>> accomplished and richly experienced lab technician"?
>>
>> Engineers don't write like that, but marketing people do.  Yep, I think
>> I'm making a safe bet.
>>
>> I was badly burned once by believing the Amazon reviews of a signal
>> processing book.  So now I read those reviews *VERY* carefully!
>>
>> [-Rick-]
> 
> Even honest reviews can be completely misleading. If the reviewer is the
> wrong audience for the material it often gets a scathing review, while
> the reviewer should probably just think "that's not for me" and move on.
> I Googled for a venison recipe at the weekend, and choose one with lots
> of 5 star recommendations and a couple of 1 stars. The 1 star reviews
> were from people who don't like the gamey taste of venison.

My control systems book has a scathing review on Amazon from some guy 
who's all bent out of shape because it doesn't cover formal robust 
control.

From his wording I'm pretty sure that it's the same guy who I had an 
email exchange with, who seemed to want to be able to go from ignorance 
of control theory to being a robust control expert with just one easy to 
read book.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com