The Secret Society of Lathe Trolls

cd4cutter · Joined: 08 Jul 2008 Posts: 36

I recently received an inquiry from Doug6N about the RIAA recording pre-emphasis. Although I responded to him, I think this topic still needs some further airing out because I'm sure there is still confusion about it. So I'm repeating most of what I said to him for everyone who's interested.

As I've said before, the RIAA pre-emphasis that is required to convert constant amplitude recording is almost never published because most cutters never cared about it. Professional cutting amps had the EQ built in as a switchable filter, and the cutting engineer simply switched it into the circuit. The only time it was switched out was when certain testing was done on setting up the cutter head. Here are the RIAA EQ curves as applied to constant amplitude recording and playback:

Curve A is for recording, curve B for playback by a constant amplitude (crystal or ceramic) cartridge. Note that this curve looks a LOT different from all the other RIAA curves that you've seen published and that there is much less level deviation from flat. That's because the commonly-seen curves are playback curves and INCLUDE the 6dB/octave de-emphasis required to convert a constant velocity magnetic pickup to constant amplitude in addition to the RIAA de-emphasis. So, if you've already got your cutter head EQ'd to do constant amplitude recording, you need add only the above curve to do the RIAA pre-emphasis.

Now, some particulars regarding the application to home and semi-pro recorders: I'm not really too familiar with the early semi-pro recorders which is what I would call the Presto, Rek-O-Kut and similar units. These were designed primarily for use by radio stations to do air checks, record local commercials, etc. before the age of magnetic tape (i.e. pre WWII). There wasn't too much standardization of EQ curves back then, even among the companies that recorded for commercial, public release records (Victor, CBS, Decca, etc.). That's why you'll find about seven EQ playback settings noted on the early hifi preamps that date from the 1950 period. RCA had been using an EQ curve that they called "New Orthophonic" in the postwar period, and they managed to get the newly-formed Audio Engineering Society to approve the curve for general commercial release use around 1947. The Recording Industry Association of America (RIAA) was also really just getting going then, and they adopted this curve for their official use as well. Refer to the RIAA curves shown above: The EQ presented by this curve had the original intended purpose of reducing rumble from the record pressings (by rolling off the bass response below 50Hz during playback - which requires a corresponding boost during recording) and reducing high frequency tracing distortion (by reducing recorded treble levels by around 12dB or so above about 2kHz - which requires a corresponding boost during playback). These EQ changes are with respect to constant amplitude recording which was the basic de-facto standard going into this time.

So, in order to understand what part of the RIAA EQ you need to implement, you need to know how your cutter behaves with no EQ. Crystal cutters are essentially constant amplitude cutters, but they have a mechanical resonance in the audio range which causes their response to fall off at higher frequencies. Magnetic cutter heads tend to have a pronounced resonance peak in their response below which they behave as constant amplitude and above which they behave as constant velocity. Most makers of cutting systems that used non-feedback magnetic cutters used some form of electronic EQ to try to make their cutters behave as constant amplitude over as much of the frequency range as possible. This usually meant that they had to EQ a notch or dip at the cutter resonance, then add some treble boost above the resonance. This EQ will be built into the maker's recording amp which matches the cutter head model, and you probably can't switch it on and off. In addition to this fixed basic cutter EQ, the amp may also have switchable RIAA (also adopted by the NAB - National Association of Broadcasters) EQ.

Long story short(er), if you want to make records that can be played by most people and sound correct on their home record players - all of which built since about 1947 are nominally designed to conform to the RIAA playback EQ, you need to implement some EQ in your recording that produces flat frequency response when played back with the RIAA de-emphasis. Figuring out academically the specific EQ that you need to add to your particular cutter is probably a waste of time due to all the variables, including the age of your particular unit. Your best bet is to attach a 10-band or higher resolution graphic equalizer to your cutter amp and use it produce the flattest frequency response you can get when playing back test cuts on a good quality stereo turntable equipped with a good modern magnetic cartridge played thru an accurate RIAA preamp. Use an audio signal generator (and AC voltmeter on playback) capable of covering the entire audio frequency range, particularly from about 30Hz to at least 15kHz. Set your recording levels to produce about -20dB from your standard loudness for these tests. You DO NOT want to record very high frequencies at levels above this for two reasons - you run a very real risk of burning out your cutter head because you will need to add a lot of treble boost to achieve flat frequency response, and even the best modern playback cartridges cannot handle high levels at high frequencies, so even if you manage to cut high levels, you cannot play them back accurately so you can't tell if your recording is correctly EQ'd. You'll need to make several tests of spot frequencies throughout the audio frequency range - record some spot tones over the whole range of frequencies, play them back into the AC voltmeter and measure the output level at each tone, adjust your recording EQ, repeat the test - until you come up with fairly flat results, i.e. the same output voltage on the voltmeter at all the test tone frequencies. It's tedious and time consuming, but it's the only way to get the result you want.
_________________
Collecting moss, phonos, and radios in the mountains of WNC

cuttercollector · Joined: 11 Jun 2006 Posts: 307 Location: San Jose, CA

All of this is filling in a few big "holes" in my knowledge.
Thanks for breaking it down for some of us who are here and trying with what knowledge we have. I have tried to make sense of the eq curves for particular Westrex stereo cutters and older mono Rek-O-Kut and Presto ones. Not to mention some of the older crystal ones. Some little home recorders use the same element for record and playback. Being as you say crystal and piezo elements are constant amplitude, what happens when an element like that is used for record and play? Do the curves complement each other to produce more or less "flat" overall response or "fight" producing a double boost or cut? This is not asking if they conform to RIAA or any other older standard. I know further eq is required for that, and also older cutters are not capable of wide response. Just does it do generally the right thing for record and playback? Treble boost and bass cut on record and the opposite on playback. I also know that the low frequency curve on piezo devices is governed by the input impedance of the amp it is hooked to. Too low rolls off the bottom end. I guess that is why ceramic pickups worked better with tube amps.

One more related question. With regard to your test procedure, some of us have tried what you recommend making a flat overall record/play response using a modern magnetic pickup and good RIAA preamp as a reference. But instead of an oscillator and volt meter checking specific frequencies, I have used a pink noise source and an RTA, adjusting record eq for flat playback response. What do you think of that idea?
Of course you don't cut at max level when you do that, just a little bit above the residual noise floor.

Thanks for helping out. You have "been there and done that".

cd4cutter · Joined: 08 Jul 2008 Posts: 36

Yes, you can use pink noise as a test source and measure your result on an RTA. But the results will be more ambiguous and imprecise because of the noisy character of the signal. This would be a good way to start your process to get the response "roughed in". But I would suggest using the sinewave tones to fine tune the response after that.

Each transducer, be it magnetic or PZT has its own characteristic frequency response. Yes, if you use the same unit for both recording and playback, the peak caused by the natural mechanical resonance will cause both a peak in the recording that you make, AND it will cause the same peak to be ADDED to the playback from that recording. So, anomalies in the transducer will be doubled when it is used both for recording and playback. That is, unless you fix the problems with EQ. You will separately need to, for example, add notch EQ at the resonance in your record amp to flatten the response while cutting. Then you will also need to notch the resonance of that same transducer in your play amp when using it for playback. Maybe the very same EQ curve will work for both recording and playback, but probably not. The transducer behaves a little differently when being used as a cutter than it does when used for playback, so the resonance will move to at least a little different frequency, and the response both lower and higher than resonance will also probably behave differently when playing as opposed to cutting. You just have to experiment to find out.

Yes, crystal and ceramic PZT transducers are high impedance devices. Particularly the newer, better ceramic cartridges have a VERY high output impedance. You are correct that they require a VERY high input impedance in the amplifier in order to avoid loss of bass - higher, in fact, than that typically presented by even a vacuum tube circuit. I suggest making your playback preamp have around 3 MEGohm input impedance when used with modern, small ceramic cartridges.

You need a good playback system to be used as a reference for adjusting frequency response. Unfortunately, NO playback cartridges have perfectly flat response. If you really want to know what's going on, you need to find some of the good professional test records that were made during the 1960s thru 80s that were designed for measuring cartridge response and use them to calibrate your high quality pickup cartridge. These records were made by Bruel & Kjaer (B&K), JVC, RCA, and CBS laboratories, along with some others. I don't believe that anyone is making these anymore, so you'll need to find old ones, preferably new old stock (NOS) that have never been played. The other problem with these records is that they rapidly wear out, especially when used with poor cartridges and styli. Frequencies above about 15kHz get wiped out pretty easily in one or two plays with poor cartridges. So you need to find NOS records or ones that you can be assured are still in good condition. I know it's discouraging, but that's the state of things these days. It's always been a tedious project, but at least the equipment and records were readily available back in the day.
_________________
Collecting moss, phonos, and radios in the mountains of WNC

Doug 6N · Joined: 21 May 2007 Posts: 66 Location: Washington

Hi CD4Cutter:

I was going to get back to you and just had'nt sat down to do it.

Aha!! I now remember one time seeing this particular curve. And thinking WTF it's wrong. LOL! Now with your explaination it's makes complete sense. And. Trying to use that published curve as the recording characteristic. No wonder it was'nt worth a darn. Somehow in all the years I've read about audio and it's quite a few. I've totally missed this. Perhaps because like you say it's never thought about. But as several of us have figured out. Something was wrong with our approach.

Thanks ever so much.

Doug

PS. Could you post a larger file of the pic. I'd like to be able to print it out larger or where did you get this from? And I'll hunt up the source.

cd4cutter · Joined: 08 Jul 2008 Posts: 36

Doug, the original article that contains the above RIAA graph was printed in the Journal of the Audio Engineering Society (JAES) July 1959, Volume 7, number 3, titled "Automatic Plotting of Cartridge Response" by C.P. Germano. Germano worked for the Clevite Brush Corporation which was one of the early makers of crystal and ceramic cartridges. This is an excellent article and one of the few that gives good detailed information about the physics and characteristics of PZT transducers - it's well worth reading if you are working with these devices. The original graph is about the same size as the scan I have posted here and isn't too good for extracting detailed information. Since I was interested in making a proper preamp for ceramic cartridges, I worked out a pretty good graphical interpretation of the information by graphically subtracting a 6dB/octave slope from the published info for the constant velocity pickups. Below is the tabular data. Note that this is not mathematically derived, but it should be accurate to within a few tenths of a dB which was good enough for what I wanted:

This table is scanned from the RS-211-D document published by the Electronic Industries Association (EIA) and represents the identical standards as adopted by the RIAA. The handwritten rightmost column shows the tabulated EQ in dB required to implement the constant amplitude (CA) RIAA playback characteristic. The printed column next to it represents the same data for constant velocity (CV) pickups. For recording, simply invert the EQ curve and negate the dB numbers (note that the leftmost column for CV recording is just that relative to the CV playback data). This should give you what you need to develop your own EQ for either recording or playback.
_________________
Collecting moss, phonos, and radios in the mountains of WNC