Author Topic: Harp Frequency Response  (Read 2654 times)

Offline pnoboy

  • Mark I
  • ****
  • Posts: 395
    • View Profile
Harp Frequency Response
« on: December 31, 2015, 01:36:10 PM »
Using a method known in the guitar-pickup industry, I measured the frequency response of the harp in my 1978 Stage 73 piano.  The technique employs a coil that is used to couple to a pickup via mutual inductance.  The coil is driven by a constant-current amplifier, and the output of the harp is sent to a high-impedance preamp followed by an integrator.  I used my oscilloscope with a 10X probe for the preamp, and did the integration mathematically in Excel.  The 10X probe provides a 10 megohm impedance with a very small amount of capacitance.  I connected various resistors across the output of the harp to see how they affected the response, and I also connected a 220pF capacitor, which would represent about 8 feet of coax cable connected directly to the harp.  The graph below shows my results.  The curve labeled "Pete" shows the simulated response of the Peterson preamp combined with the 42.2k resistor.  The Peterson preamp had a 33k input resistor, but the 42.2k was the closest value I tested.  I should note that putting resistors across the harp reduced the output, but I normalized all graphs to have a 0 dB output at 400 Hz so their frequency responses could be more easily compared.  I would estimate the accuracy of my measurements to be about +/- 1 dB.

After doing this test, I also did a listening test.  This test used a solid-state amp whose tone controls were set to a flat response, combined with a hifi speaker with a 40 Hz to 20 kHz bandwidth.  The amp had a 1meg input resistance.  Using the same resistance values as in the previous test I listened as hard as I could to detect any changes in tone.  Even though the graphs show response differences up to the highest impedance, I could not detect aural differences for any resistance between 50 kohms and 1 megohms.  I believe this result is caused by the fact that the Rhodes has little response beyond about 8 kHz, and even at this frequency, the only response is the sound of the hammer hitting the tine, which damps out very quickly indeed.  Except for the initial sound, the Rhodes produces little output over 3 kHz.

The stage models used a 10k pot for the volume control, and it is clear both from the graphs and my listening tests that this value reduced the initial ping or tinkle associated with the Rhodes.  Whether or not that's a good thing depends on your personal taste.  The Peterson preamp was designed to eliminate this initial attack sound as it substantially reduces response above 3 kHz.

Based on the response and listening tests, my conclusion is that any input impedance higher than 50 kohms to 100kohms will capture the full Rhodes sound.
« Last Edit: December 31, 2015, 01:39:28 PM by pnoboy »

Offline Tim W

  • Sparkletop
  • ***
  • Posts: 80
    • View Profile
    • Retrolinear Inc.
Re: Harp Frequency Response
« Reply #1 on: December 31, 2015, 03:37:02 PM »
Well done!

Can you plot frequency (x axis) on a log scale?  The curves will probably look more familiar and easier to read/understand. Most audio plots are loglog (both x and y) but you already did the log conversion for the y axis by calculating dB. If excel is going to be a pain, you can send me your data and I'll dump it in Matlab.

Best
Tim


Offline pnoboy

  • Mark I
  • ****
  • Posts: 395
    • View Profile
Re: Harp Frequency Response
« Reply #2 on: December 31, 2015, 05:16:36 PM »
No problem--see below.  I had thought about plotting loglog, but using a linear frequency scale allowed greater detail in the area where the roll offs occurred.  I restricted the frequency range to help improve detail in that area.
« Last Edit: December 31, 2015, 05:24:05 PM by pnoboy »

Offline The Real MC

  • Mark II
  • *****
  • Posts: 751
    • View Profile
    • My '67 sparkletop Rhodes
Re: Harp Frequency Response
« Reply #3 on: December 31, 2015, 09:32:32 PM »
Very interesting.

Smart move using a 10X probe.  The Rhodes pickup is very sensitive to loads, even with 1M impedance load. 

They changed the pickup design over the years, so you tested the later ones.  I bought a 1967 sparkletop that was missing the preamp and found that the best sound was to take the signal right off the harp into a DI with 10M impedance.  Tried that same trick with my 1976 stage piano which bypasses the 10K volume pot and it still sounded dull and lifeless compared to the sparkletop.  There is a tone difference with pickups of different eras.

I have also found that the early pickups can saturate.  Position it too close and you hear a "thunk".  I exploited the "saturated" feature by positioning the pickup such that I got a pleasant transient "thunk" with harder playing, which didn't work with later pickups.  Rhodes likely changed the design to get cleaner more powerful output and better S/N ratio.


Offline The Real MC

  • Mark II
  • *****
  • Posts: 751
    • View Profile
    • My '67 sparkletop Rhodes
Re: Harp Frequency Response
« Reply #4 on: December 31, 2015, 09:38:49 PM »
The selected impedance and frequency response of the stage pianos and peterson/hagler preamps was intentional; they mask the noise of the pickups.

If you take the signal directly off the harp and use the wrong preamp or DI, you'll hear the inherent noise.  Use the right DI (like a Countryman Type 10) and the noise disappears while retaining the original tone.

Offline pnoboy

  • Mark I
  • ****
  • Posts: 395
    • View Profile
Re: Harp Frequency Response
« Reply #5 on: January 01, 2016, 06:26:52 PM »
If the noise disappears with a particular preamp, then the conclusion must be that the noise was caused by the preamp(s), and not the pickups.  I would expect that the only noise the pickups would produce would be Johnson noise, due to their resistance, and line noise or other EMI that gets in.

Offline Ben Bove

  • Vendor
  • MIDI Mark V
  • *****
  • Posts: 3279
  • Formerly bjammerz
    • View Profile
    • Retro Rentals Website
Re: Harp Frequency Response
« Reply #6 on: January 04, 2016, 01:29:34 PM »
Great stuff pno!  One thing I was really looking to research, was comparing the frequency response of a harp against the output of the factory speakers.  Reason being, I often get complaints that "the Rhodes sounds better in the headphones and muddy in the speakers" or "what I hear in the headphones is brighter than the suitcase cabinet," which also has led a lot of people to attempt to add bright kits to their suitcases with tweeters etc. but is more of a crude hit an miss on speakers they throw in there (probably far out of the frequency range that should be).  I can also say to my ears, that they are correct - direct headphones vs. speaker cabinets have an audible difference.  I'm sure there's some acoustics in play with the speakers being at your feet and what is deflected up to your ears, but I'd be interested in how things test out.

Is there any way to compare the charting of a Peterson output, with the output range of a 12" speaker from misc manufacturers like Utah, Jensen etc?

« Last Edit: January 04, 2016, 01:43:01 PM by bjammerz »
Retro Rentals
Vintage Music Gear

http://www.RetroRentals.net
(818) 806-9606
info@retrorentals.net

FB: https://www.facebook.com/retrorentals.net/
IG: @RetroRentalsNet

Offline pianotuner steveo

  • MIDI Mark V
  • *****
  • Posts: 3290
  • A keyboard player in love with vintage guitars!!!
    • View Profile
Re: Harp Frequency Response
« Reply #7 on: January 04, 2016, 01:54:28 PM »
Just a wild guess here, but maybe the designers picked speakers based on the straight frequencies produced by the tines and not considering overtones, partials, etc.

If you go by that , then the frequency response of an 88 key keyboard would be 27.5hz for low A, up to just under 6k for high C.

I don't think the factory speakers in a Rhodes went up to 6k efficiently which could explain why it sounded brighter through headphones. They are basically woofers (?)

They were likely more concerned that the speakers could handle the low bass.

Just a thought.
1960 Wurlitzer model 700 EP
1968 Gibson G101 Combo organ
1975 Rhodes Piano Bass
1979 Wurlitzer 206A EP
1980 Wurlitzer 200A EP
1980 Wurlitzer 270 Butterfly Grand
2000 Yamaha acoustic piano
2004 Hammond XK3
2009 73A Rhodes Mark 7
2009 Korg SV-1 73
....and a few guitars...

Offline Ben Bove

  • Vendor
  • MIDI Mark V
  • *****
  • Posts: 3279
  • Formerly bjammerz
    • View Profile
    • Retro Rentals Website
Re: Harp Frequency Response
« Reply #8 on: January 04, 2016, 01:58:59 PM »
I'd definitely agree on ulterior reasons for speaker choice - one of them being confirmed that a major concern was the weight of the speakers to be lightweight in order to prevent speaker baffles breaking in product shipment.  So that sort of points to manufacturing concern rather than the ideal frequency response for a Rhodes.
Retro Rentals
Vintage Music Gear

http://www.RetroRentals.net
(818) 806-9606
info@retrorentals.net

FB: https://www.facebook.com/retrorentals.net/
IG: @RetroRentalsNet

Offline The Real MC

  • Mark II
  • *****
  • Posts: 751
    • View Profile
    • My '67 sparkletop Rhodes
Re: Harp Frequency Response
« Reply #9 on: January 04, 2016, 08:18:13 PM »
Rhodes were built by a guitar company.  Guitar amps have a max upper frequency of ~5K and they made 100x more amps than pianos.  I don't think they could justify a speaker that could go much higher than that.  Exception might be the JBL "F" speakers which were optional.

Yes headphones are going to sound different than speakers in a cabinet.  Few speaker cabinets have the flat frequency response of good studio headphones.  In a non-ideal acoustic environment, all bets are off as the room is going to further color the sound.

Offline pnoboy

  • Mark I
  • ****
  • Posts: 395
    • View Profile
Re: Harp Frequency Response
« Reply #10 on: January 05, 2016, 06:36:42 AM »
I used to have a '69 suitcase piano--it sounded very dull.  I found out that the speakers Fender used that year were the same ones they used in the Bassman amps.  Effectively, they were woofers, and further tests proved that they had very poor treble response.  I ended up taking those four 4-ohm speakers out and replacing them with 2 8-ohm guitar speakers facing the audience, and left the other two holes empty.  The piano sounded much better.

Offline pnoboy

  • Mark I
  • ****
  • Posts: 395
    • View Profile
Re: Harp Frequency Response
« Reply #11 on: January 05, 2016, 06:45:12 AM »
Great stuff pno!  One thing I was really looking to research, was comparing the frequency response of a harp against the output of the factory speakers.  Reason being, I often get complaints that "the Rhodes sounds better in the headphones and muddy in the speakers" or "what I hear in the headphones is brighter than the suitcase cabinet," which also has led a lot of people to attempt to add bright kits to their suitcases with tweeters etc. but is more of a crude hit an miss on speakers they throw in there (probably far out of the frequency range that should be).  I can also say to my ears, that they are correct - direct headphones vs. speaker cabinets have an audible difference.  I'm sure there's some acoustics in play with the speakers being at your feet and what is deflected up to your ears, but I'd be interested in how things test out.

Is there any way to compare the charting of a Peterson output, with the output range of a 12" speaker from misc manufacturers like Utah, Jensen etc?

Sure, but there are some difficulties.  The simple approach would be to use the published response curve provided by the manufacturer, but most of the time those curves give on-axis frequency response in an infinite baffle, and not the total-power frequency response.  Additionally, the response curve of the speaker is influenced by the size and type of enclosure it's in--is it open-backed? sealed? ported? The dimensions of the baffle affect its response as does the position of the speaker relative to reflective surfaces such as the floor or a wall.  Having said all that, I think a good proportion of guitar speakers are a pretty good match for 73-key Rhodes.  I also think playing through an equalizer can be very helpful for tweaking the sound.