Old Siemens meter?

[audiox]

you could try to get hold of some transistorized versions of the meter amps.

I have never seen a transistorised version of the meter amp (except made by some broadcasters to replace the old tube units that were no longer available). The reason is that transistor circuit is so small that it can be placed in the meter chassis. The first generation was a valve amplifier + external moving coil meter. The second generation was all-in-one transistor PPM (I believe NTP was one of the first in the market).

 

[Michael Tibes]

Well, I've seen loads, all collecting dust. There is definitely one from the 2xx Sitral series, I have 2...

Michael

 

[mad.ax]

U270, 271, or 370 cost near as much as U70 or 71... We are talking ~300€ for a pair, wich is not very expensive per se,
....but where would be the DIY fun?

Plus, once I will have a working PPM board I will finaly be able to use all those pretty bakelite milliampmeters that I have accumulated for years!

Axel

 

[PRR]

> There's indeed a lot of similarities between your plan and the original one!

Only a little bit. Indeed the difference between a 5 minute brain-burp and a 5-month development process.

Naturally they filled the input with switches and pads. Amplifier, supersonic filter.

I'm startled the EBF89 can supply its own AVC voltage. Its plate output drives transformer and vacuum diodes, charges C9 C10 negative. Large input signals force the EBF89 "off", reducing the 55dB range to the ~~20dB which fits on a simple mechanical meter.

I don't think all the crystal diodes do much. Probably damp or clamp excess voltages.

The "0dB at 1.4mA, -50dB at 3.4mA" made me think they used the AVC tube's plate current directly.

No, they read the AVC voltage in tube 4. Which really does the same thing as reading EBF89 current, except they have a trick peak/average network parts 9 10 19 28 29. Peak and average are combined from both plates. R30 trims the no-signal current to a bit over 3.4mA. Signals turn-off tube 4, current falls to 1.4mA, 1mA.

Either way, a STRONG signal can't burn-up the meter. Meter gets the most current for a zero-signal condition. This current will be set when tubes are installed, and informally checked every time there is a pause in the program (if the meter does not go low when the program is quiet, the operator may notice).

That's a nasty plan to copy without a radio-station's budget. The switching could be simplified, the 150C2 can be a Zener string, tube 1 could be op-amped, the lamp-bridge is clever and convenient but not essential, but tubes 2 and 4 are very hollow-state techniques, fairly complex, full of trimmers.

What do you want? A true -55dB...+5dB PPM? Or a dancing light show? Some back-bias and a transistor should make it dance the top 20dB of program like a VU meter. The numbers won't be right, the ballistics won't show either peak or loudness, but it might keep your clients mesmerized.

 

[PRR]

This is the actual scale of the Siemens meter. At this scale it reads 106 pixels per milliAmp, that's how I estimated the dead-rest (no current) position.

 

[PRR]

This is the simplest PPM I ever saw. Power is classic Neve +24V. The top is just a utility booster, in case you need to meter weak signals. The bottom is the actual PPM. There's some really nifty tricks in there. It has outputs for 1mA and 0.5mA meters; it is obvious how to rig other sensitivities, though meters over 10mA may cause the decay rate to fail BBC testing.

 

[PRR]

Here is a simple hack to make the meter idle at 3.45mA, and go to 1.44mA at nominal "100% modulation". I think the 0dB and -40dB marks will be right. -20dB actual level will hit the -15dB mark on the Siemens: the Neve is pretty good linear-in-dB and the Siemens has the dB crunched up at small signals. It will swing toward +5dB mark but won't be exactly 5dB change of level. So as a soft-medium-loud indicator, it is useful over a full musical range; if you need exact numbers, use a real meter.

 

[audiox]

Another example of very simple PPM circuit is PPM driver module made by NTP (type M-900). It is same size as NTP M-100 op-amp. It has unbalanced input for audio and output for a moving coil meter.

I don't have the internal schematic of it but it can't be very complex because of the space limitations (25x30x17mm). The module is designed in early 70s and made of discrete transistors. Despite the small size and simple circuit, specifications are very good according to the datasheet.

 

[mad.ax]

Wow PRR! Thanks for your detailled answers!

I agree that copying a U70 would be a nasty plan, but comparing how they did it in 1956 (U70), in 1969 (Neve 192), or in 1976 (Studer 069) is definitely instructive....

>"What do you want? A true -55dB...+5dB PPM? Or a dancing light show?"

I will have to ask this question to my friend, but I guess he will answer something in between...
For a true PPM, nowadays, and for 90% of applications, nothing beats the cost/precision ratio of a computer, and honnestly there's something quite appealing in a vintage "dancing light show"!
So I guess if the light show could be not totaly useless, my friend will be happy!

I will definitely breadboard the Neve circuit, if only to see how useful it is!
As you said it's pretty good linear-in-dB, redraw the scale on the Siemens is an easy option.

Off topic: All the pictures that you post are displayed twice when I read the forum, but only appear once in the topic summary when I reply....  Is that a problem with my linux mozilla browser? Seems to happen only with YOUR posts though...

Axel

 

[audiox]

Neve circuit is designed for BBC scale which is "dB linear". The scale is from -12dBu to +12 dBu (or 1 to 7 in the actual scale markings). Total range is only 24 dB.

Studer circuit is designed for "dB linear" scale from -40dB to +5dB (where 0dB is +6dBu). Total range is 45 dB.

None of the above fit to German scale (DIN 45406) which is "compressed" at the low end (not dB linear except the upper 1/3). Total range is 55 dB. That kind of circuit can be found for example in NTP 177-400 peak programme meter. That NTP circuit can be swithed between "db linear" or DIN scale just by changing a few component values.

 

[Michael Tibes]

Do you have a U270 schemo? If it's not on Kubis site, I could try to dig it up. Maybe that's easier to rebuild?Reconstructing the U70s seems a bit like a waste of effort just for a meter, you could build something very nice which belongs in the signal chain instead.

Anyway, I don't wanna spoil your DIY fun 

300 € seems way too much for a pair of U 270, but I don't have any around which are for sale unfortunately. I certainly paid a lot less (way less including the meters) - and those are modules which normally noone has a use for, so why should the price rise? I believe some people would be happy to let them go for little money. Just gotta find one...

Michael

 

[audiox]

I did some rough simulations of the Neve PPM circuit (the part after the first two transistors which are only a gain stage). It works fine except the diode/resistor circuit that approximates the log curve in parts. The accurary of the log conversion is very poor. It must be highly dependent of the diode types. I assumed the MS2027 diode to be 2.7V 500mW (just a guess). Any ideas?

 

[mad.ax]

>Audiox,
if the Neve circuit is indeed limited to a 24 dB range, then I guess it cannot remain a good candidate.

The 45 dB range of the Studer looks less far from the goal, and making a new scale to match it would be easy

The NTP may be perfect, but I couldn't manage to find any schematic for it...

>Michael
if I understand the numbering scheme correctly, the U270 would have been made by Siemens, and the U370 would be the same thing, but built by TAB?
There's a whole manual for the U370 on Kubi's site, and below is the schematic.... Definitely something easier to buy than to DIY, as I don't see myself sourcing 2 transformers and a coil, just for a PPM!

Axel

 

[PRR]

> pictures that you post are displayed twice when I read the forum, but only appear once in the topic summary when I reply....  Is that a problem with my linux mozilla browser?

Huh. You are right.

When I use MSIE (7 or 8), and Attach an image, I see a text-link to the image, but NO image.

That's on this forum. On another forum with the same forum-software but different options, attached images appear as thumbnails, which may be expanded.

When I want the image to appear "in" the text, here, I copy the image URL to IMG image tags, forcing it to appear.

BUT.... in FireFox/Windows, I _am_ seeing the image where the attach-link is, plus (when I do the trick to force MSIE to display), where I put the IMG tags.

Very odd. I did not realize that MSIE and FF were displaying differently.

There's no attached-images when not logged-in. Presumably Ethan has blocked access to attachments by unknown visitors.

I believe reply-review page is supposed to be without-images; some confusion there.

Your images display different when they come from another server; some of your very-large images are too big to be Attached here.

FF is also refusing to take the Alt-S shortcut to Submit/Save a post/edit.

 

[PRR]

> Neve circuit .... Total range is only 24 dB.

My partial simulation shows reasonably good V/dB linearity over 40dB of input. Dot-line is perfect, solid line is simulated from partial schematic, notably omitting the rectifier.

(How does that display?)

> the log conversion is very poor.

Hmmm, we disagree. Wonder what the difference is.

> I assumed the MS2027 diode to be 2.7V 500mW

My sim only has a small 5.6V Zener, I edited the model for 2.7V, and let it run. I am aware that a very low-volt Zener is "different", but I didn't think it would be very important?

Clearly Neve was able to get these things sold and paid-for. The BBC PPM is a broad-zone indicator, not high resolution, so maybe "poor" was good enough? Or maybe he painted the dials to match his poor conversion?

> DIN 45406... is "compressed" at the low end

Yes. A nice scale, but hard to do well. Since we have digital overload indicators, and usually good monitoring (broadcast engineers occasionally work "deaf"), do we need really great PPMs? Or do we just want an analog distraction from dull digital machinery?

> redraw the scale on the Siemens

_I_ love the 1950s graphic. _I_ would leave it original, just make a mental note that it isn't dB-perfect.

What is a meter for? To know when you clip; but that's why you have the OverLoad LED on your other gear. To know when the mike went dead; any meter will give a zero indication. To know the general average loudness: soft for the flower show, loud for the football game; just remember what part of the scale seems to be right loudness for several types of program.

If you -need- a -37dB level to tune a compressor, you have DVMs and maybe a 40-segment LED meter somewhere, which can give you 0.011V on the nose. While the 1960s broadcast engineers may have used these PPMs to check -37dB levels, the limiter factory would have used a gain-set with a good voltmeter and precision attenuator to get reference tone.

> I don't see myself sourcing 2 transformers and a coil

Most of that U370 plan is German thoroughness. Added input switching, supersonic filter, killer rectifier driver. Then "Platte U370a6" is the same idea as Neve's Zener-string. Neve stacked them up and let the errors average-out, TAB gives six separate break-point trimmers and paid some apprentice to adjust every one. "Platte U370a4" is a very high precision voltage to current converter, where Neve just buffered to a 10K resistor.

One difference: Neve log-converts the rectified audio, then the attack/decay filter; TAB rectifies onto the attack/decay filter and then log-converts. This affects your choice of time-constants, but is probably all the same to the user.

 

[audiox]

My results are quite different. When I have time I will investigate what is the problem.

Interesting that the linearity is that good over 40dB because only 24dB is needed (the BBC scale is only 24dB. "1" is -12dBu and "7" is +12dBu).

The positions of the 27K and 8K2 resistors in your simulations are different from the original schematic.

 

[mad.ax]

About the pic attachment...
Your last picture (hosted on tinypic) display well when browsing, and also while answering.
While typing an answer, I too only see a text link to the image, no matter if I use the attach function (prodigy hosting), or an external url.
I have to hit the post button and read the thread to check if it's displayed correctly (that's why I got an answer in this very thread before even asking my question!)
That's probably something to be asked to Ethan...

Back on the PPM topic...
I'm ordering some zeners and will breadboard the Neve circuit. This way, I will see if it's useable enough...
I love the 1950's graphic too! It's an important part of the 'analog distraction". If I edit the scale, I will use an high res scan in the GIMP in order to keep the look intact.

Axel

 

[audiox]

The NTP may be perfect, but I couldn't manage to find any schematic for it...

I have schematic of one NTP meter amp which is designed for the DIN scale. PM your e-mail and I will post it to you.

 

[audiox]

I tried to find error in my simulations but had no succes. Because I had some spare time I built a prototype and made measurements. The results were very close to my simulations.

The gain stage built aroud the first two transistors was not included in my proto because it just provides gain of aproximately 15 dB. So the generator is connected directly to the 1K resistor.

The log converter is quite accurate in input level range from -12dBu to +6dBu. Range from -15dBu to +9dBu requires slight modification of the meter scale at both ends. Outside that the error increases fast.

So the usefull range of the Neve B192 PPM circuit is 24dB that is the same that the BBC scale requires. Not a surprise. For other scales like DIN or Nordic this circuit can't be used.

Frequency response is quite good. There is no difference between 1 kHz and 20 kHz above -9 dBu. At -15 dBu the difference is aproximately 1 dB.

 

[PRR]

> The positions of the 27K and 8K2 resistors in your simulations are different from the original schematic.

?? You are correct. I don't know why I did that. It was several years ago and I do not remember. If I simply made a mistake, then why is my simulated linearity so good? Maybe Neve made a mistake on their drawing? But I am not smart enough to see a bad plot and deduce that the resistors were drawn one-up from where they should be; that's a fairly subtle insight.

> I built a prototype

Great! I never trust the computer. Even when, as in this case, it told you what you measured.

Is it possible that "my mistake" is correct, or useful, giving tolerable linearity beyond the 24db range the BBC requires?

> Frequency response ...no difference ...above -9 dBu. At -15 dBu ...1 dB.

I didn't think on this but it is no surprise. Elegant log converters have frequency response proportional to level. This switch-point approximation is near 1Meg for small signal and a few K for large signal... a hundred pFd of stray would explain your measured response.

The rectifier in the U370 is crazy. It appears to kick-up 100V output level. That makes sense: the simple diode has 0.6V drop, even biased it will have a couple tenths of a volt offse, and 54dB above that is 100V. That's why the rectifier driver looks like a power amplifier.