Sparky

Just wondering why you chose IIR filtering vs. FIR in this application?

Camclone,

My comments were not intended to say that you and your station were non professional in scope, rather from the limited information provided I had no way of knowing this, and thought perhaps it was of your own design.

As a +30 year veteran in and out of the broadcast business and a wireless design engineer here are a few points to consider.

Receiver lock. Yes as you have pointed out, different brands of receivers have varying levels of engineering quality built into them. Not all brands follow the nominal design guidelines needed to receive standardized broadcast transmissions. Because of this, some people, due to their individual choice of receiver brand purchased, will have a less than satisfactory listening experience. There is nothing you can do to prevent this except to adhere to the broadcast transmission standards of your country as closely as possible such that the largest majority of listeners who do have adequate receivers will receive your signals properly.

Loss of lock in FM receivers as you describe or have been seeing is not due to a loss of main FM carrier PLL lock. All superhetrodyne receiver designs use an unmodulated local oscillator to convert down the FM signal of choice to a 10.7MHz intermediate frequency. This frequency is then sharply filtered and applied to the modulation recovery circuits (demodulator). The quality of the IF filtering as Leif pointed out will have a significant impact on the quality of the recovered audio and subcarrier signals. All FM broadcast receiver ceramic filters must be engineered to have a sufficiently wide and flat bandpass response, in combination with very low group delay for all of the allotted FM channel bandwidth (+/- 100kHz). Typical ceramic IF filters for FM broadcast reception use anywhere from 150-330kHz of bandwidth depending upon the designers ultimate design goals. In most instances the loss of "PLL lock" you describe can be largely due to issues within the receiver stereo demodulator sub-circuits. A large part of distortion from these circuits come from non-linearity in the phase discriminators, subcarrier filtering, and internally generated synchronous AM of the 19kHz stereo pilot. Signal multipath distortion compounds the problem of synchronous AM of the 19kHz stereo pilot.

However, there are many reasons why a loss of "stereo PLL lock" in a high quality receiver is encountered that are directly caused by the FM station itself.
These are: synchronous AM of the FM carrier and MPX subcarriers, transmitter interstage group delay, MPX tilt at the exciter, passband tilt of the antenna array.
In all my experience in the broadcast engineering business, a lack of symmetrical interstage or amplifier group delay is the most common of the above problems. Many times I personally have seen what this can do to a station’s 100kW FM signal.

In an ideal FM transmitter the main carrier deviates symmetrically from center frequency regardless of modulating frequency. This in turn generates modulation sidebands with equal amounts of energy. As these sidebands propagate through various amplifier circuits, each frequency group will pass at a slightly different rate such that all groups are not delayed equally in time. However the design goal of a transmitter engineer is to build symmetrical group delay throughout the system which equates to constant group delay or linear phase shift with modulating frequency. But many times the interstage coupling between the exciter to the IPA or the IPA to the final amplifiers introduces significant group delay causing anything from tiny to massive amounts of distortion seen only at the receiver. Throw in signal multipath distortion it’s any wonder a receiver can faithfully recover any listenable audio. Excessive processing with uncontrolled clipping makes a bad situation that much worse. So if the transmitter modulation group delay is large, the MPX signal will undergo lots of phase shifts that will make the received audio and MPX subcarriers appear to "lose lock". You the station engineer need to perform a sweep analysis of your entire transmitter chain (from exciter all the way through to the antenna) to see if this is indeed the case. If the results of your analysis shown everything is perfect, then you have done your job to ensure the station is transmitting a clean signal. The rest is up to the listener and the uncontrollable effects of how your signal propagates around the coverage area. Can’t do much more than that.

From a design engineering perspective, multi-module MOSFET amplifiers are very much prone to the effects of group delay distortion. Because each amplifier module splits the incoming signal, amplifies it, then recombines the outputs to create the TPO, any slight coupling mismatch will create noticeable group delay, both at the carrier level and MPX modulation sidebands.

What a bizarre request… but having read the post a few times I think I understand what his issue is.

Camclone, it appears you have a poorly designed (or broken) PLL synthesizer in your transmitter, specifically the loop filter.
This loop filter is a group of circuits that generates the feedback control voltage keeping the transmitter on frequency.
If the loop filter circuit time constants are improperly chosen or wrong, you will have high instability from audio modulation, especially lower bass frequencies. Changing Breakaway’s frequency equalization to accommodate the faulty transmitter is not the way to fix this problem. To do so would be similar to asking a mechanic to remove the doors off your new car because the air conditioner is broken. 😉

Is your transmitter design a kit and was assembled by you? If so, I suggest taking a close look at the component values in this circuit. An incorrect resistor or capacitor value will have a dramatic effect on how well it works. If it’s of your own creation, then I suggest you find more information on loop filter design and try again. Once you have evaluated your circuit I recommend you use the 30Hz burst tone tool found in Breakaway’s test tone settings window. This will help aid in determining if your changes were effective or not. If you have access to one, an oscilloscope is a great tool to view the loop filter output when the 30Hz tone burst is applied. It will quickly show you how well the PLL is working and it’s stability (or lack of) to a low frequency impulse. If the viewed waveform is wildly swing around each time the tone burst hits, the loop filter still needs work. However if you only see a tiny "bump" in an otherwise flat trace then you probably have it right (or very close).

Looks really good 😀 😀
However…

Plain o’l white noise is great for checking the MPX portion of the airchain. Your MPX tool demonstrates this very well, but once the signal is passed on through to an exciter the ideal world abruptly meets reality. Not all exciters are alike. Each have their strengths and limitations. Many introduce sonic artifacts that can are caused by modulator non linearity, high Q circuits exhibiting group delay, and coupling asymmetry in cascaded amplifier stages. I know many exciter designs use high frequency predistortion techniques that try to compensate for some of the non linearity, but many times it fixes one problem by creating another. Offering both options. i.e White and FM Fuchsia noise spectra will give the user that much better tool set to test their system with (IMHO).

I understand your argument regarding the 30Hz options and I agree in part. Any good designer adheres to the KISS principle (Keep It Super Simple). But the main objective of the 30Hz tool is to see how bad the overshoots really are. Remember this should be an "unprocessed" signal. The 30 Hz tone w/ burst option is to see how the exciter AFC circuits handle this kind of impulse energy. A good way to think of it is testing how well your cars’ suspension responds to an impulse like a deep pot hole. A well designed system that is critically damped the driver hardly notices the bump. Over damped and the driver gets kidney damage. Under damped and the car goes bouncing merrily along for a few km well after the pot hole event (ie. cars with no shocks… just springs ).

Exciter PLL control loop filters behave no different. Loop filter dynamics can sonically alter the sound of deep, booty style bass, or heavy bass drum beats. Under damped systems (fast PLL lock times) can create huge RF carrier overshoots that will exceed government regulations, or generate bad asymmetric carrier deviations. Over damped and the base tones get stripped down leaving no low end dynamics. The 30Hz tool is a convenient way to see how your exciter handles such extremes and distortions (if any) it creates. I know many of your customers are or will be future radio pirates that will eagerly embrace BBA as a lower cost way of sounding competitive like their professional FM counterparts. A lot of them will use hand built equipment. Why not give them another tool to help keep the gov radio police at bay? Perhaps as a compromise the 30Hz burst is shaped to help soften the initial overshoot vs. the first 1/2 cycle going right to 100% amplitude. Shaping is more like real world anyway.

Any more thought of using touch tone signal pairs for the preset control mechanism? If you want I could generate a few tone files for you to test FFT processing. Let me know.

FYI.
Attached are two photos of the MPX composite signal with white noise. The noise was generated using Cool Edit (mono 16-bit 44.1kHz) driving the BBP pipeline.

First is the direct composite output (w/ 2 SCA’s unmodulated). The L+R and L-R matrix is clearly visible with the pre-emphasis slope on each. If a BBP user has an o-scope with FFT capability then white noise modulation will aid in showing tilt when using a sound card for MPX generation.
[attachment=1:2ogu8chf]White Noise Modulation Of MPX Composite Signal With SCAs.jpg[/attachment:2ogu8chf]

The second photo is the actual RF spectra (no SCA’s). Notice how symmetrical the sidebands are relative to center (the pilot and sideband are the two discrete spikes). This is a properly tuned system that has sufficient bandwidth. Any tuning mismatches will show a noticeable skew in the symmetry. Unfortunately I don’t have anything handy to de-tune it enough to show this.
[attachment=0:2ogu8chf]RF Spectra White Noise Composite Modulation.jpg[/attachment:2ogu8chf]