Chris,
Shifting of resonance peaks during vibratory stress relief is caused three different ways:
1. As you described, a warm up of a vibrator that is NOT servo controlled, which will cause a drift up in speed with constant voltage input. This gives the illusion that the peak has shifted, when it has not. (NOT a true shift, but an anamoly of instrumentation.)
2. A growth of the resonance peak, which causes increased loading of the extreme fiber of the workpiece. Although a weak response (a doubling of the resonance peak will normally cause only a 1 - 2 % drop in the resonance) this can be seen with the right instrumentation. Scanning thru the speed range SLOWLY is a must (like 10 RPM/sec) and plotting the acceleration vs. RPM (from a tach, not back-emf of the motor.) This phenomenon was reported in a paper published by the ASM, that has been posted at:
http://www.msnusers.com/VSRUsersResearch/Documents/ASM%2004%2D2005%2EpdfThe equipment used to make the series of charts at the end of the paper had a resolution of + / - 0.03% vibrator speed regulation, and the vibrator's unbalance was adjustable over a 20:1 range. If the equipment you try to see effect with only has 2 - 5% regulation, which is what you would expect from a DC vibrator controlled with only a Variac (variable transformer), then you wouldn't see this effect. [I don't care whether you used AC or DC, (although we much prefer brushless DC, best of both worlds, tight servo, no brushes or commuator to arc, wear out, fail, cause drag, generate carbon dust, etc., etc.), but the vibration data will be FAR MORE DETAILED, if you servo control the vibrator. Metal structures have resonances that are often amazingly strong, but very narrow, i.e., they have high mechanical Q's.)
3. A change in shape of the workpiece, which does occur at times.
We have used vibration tables that show both responses 2 and 3, but the resonance of the table, not likely to be assocated with any residual stresses, remains exactly the same. (I will try to post a chart showing this on the msn/VSRUsers site, for your viewing, and will post the link here.)
By using very tightly controlled vibration in the range of 0.03%, with the ability to vary the unbalance over a wide range (20:1), and going to much higher speeds (8KRPM) as opposed to 3.4 KRPM and having a much stiffer vibrator mounting system and the ability to re-orient the vibrator without having to use an angle plate, we have pushed the envelope. We can now distinguish between an old casting (whose resonance pattern does not change) and a "green" one, with this technology. We also can evaluate whether it would be benefical to stress relieve not only before rough machining, but also afterwards.
You can see more at our website:
http://vsrtechnology.netBruce Klauba