radiomuseum.org
Please click your language flag. Bitte Sprachflagge klicken.

marantz: FM De-emphasis correction for bright high frequency response

Moderators:
Martin Renz Ernst Erb Vincent de Franco Martin Bösch Mark Hippenstiel Bernhard Nagel Dietmar Rudolph Otmar Jung Heribert Jung Eilert Menke 
 
Please click the blue info button to read more about this page.
Forum » Radios and other type of sets (Physics) etc. » MODELS DISPLAYED » marantz: FM De-emphasis correction for bright high frequency response
           
John Matheson
 
 
AUS  Articles: 4
Schem.:
Pict.: 0
27.Mar.21 00:23
Reply  |  You aren't logged in. (Guest)   1

I have just rebuilt a Marantz 2216 receiver. After the rebuild I noted the receiver's FM reception sounded 'bright'.

The FM stereo demodulator is a Hitachi HA1156W, for which I found the original datasheet and application notes online. The recommended de-emphasis is achieved by loading the channel outputs with a resistor and capacitor in parallel shunted to audio reference (i.e. the (+) or (-) rail).

For for 50μS the recommended load is 3k9Ω in parallel with 0.012μF, as shown in the application notes.

For some unnknown reason Marantz used 0.0039μF capacitors in the 2216 deemphasis circuit, which produces a time constant of 15μS, not the 50μS required for most countries or the 75μS for USA. This results in a rising frequency response and bright sound.

Not having any 0.012μF capacitors to hand, I shunted C311 & C312 with 0.0082μF capacitors on the underside of the circuit board. This corrects the high frequency response.

This article was edited 27.Mar.21 06:50 by John Matheson .

Marc Gianella
 
 
CH  Articles: 295
Schem.: 603
Pict.: 360
27.Mar.21 11:21

Count of Thanks: 1
Reply  |  You aren't logged in. (Guest)   2

I can hardly believe that.

In my opinion, the determining parts for the deemphasis are R313-C321 resp. R314-C322 in conjunction with the 19/38kHz low pass filter.

I'd check C301 and C323/CC324 and the rest of all electrolytic caps in the signal path.

John Matheson
 
 
AUS  Articles: 4
Schem.:
Pict.: 0
27.Mar.21 17:57
Reply  |  You aren't logged in. (Guest)   3

Thankyou for your reply and suggestions.

All the electrolytics in the 2216 subject unit had been replaced with new Nichicon Fine Gold capacitors tested for C value, Vloss and ESR.

Your suggested R313-C321 and R314-C322 filters do not have the appropriate time constant for FM de-emphasis.

In the 2216 there are two seperate notch filters for 19kHz and 38kHz, formed by L301/C313/C315 and L302/C314/C316 plus L303/C317/C319 and L304/C318/C320 respectively, which may work in conjuction with low pass filters R313-C321 and R314-C322 to reduce 19kHz, 38kHz (and 76kHz from the oscillator used by the circuit to generate the lower frequencies by devision) to an acceptable level in the demodulated audio.

The Hitachi application scematic shows the deemphasis components R101/C101 and R102/C102, which in the Marantz circuit correspond to R311/C311 and R312/C312 respectively.

Adjusting C311 and C312 to 0.012μF as per the Hitachi application circuit corrects the measured demodulated frequency response of the FM output, at least for the 2216 that I measured on my bench.

HA1156W Standard Circuit

 

 

 

Attachments:

This article was edited 27.Mar.21 18:12 by John Matheson .

Marc Gianella
 
 
CH  Articles: 295
Schem.: 603
Pict.: 360
28.Mar.21 21:20

Count of Thanks: 2
Reply  |  You aren't logged in. (Guest)   4

I was not quite clear with my statement.
I think the parts values in the schematic do have exactly the right values to reach the appropriate time constant in sum of the whole circuitry. R311-C311 and R313-C321 resp. R312-C312 and R314-C322 are the deemphasis circuits. The sum of both time constants makes around 42us. Imo the rest comes out from the interactions of the entire circuitry.


What are the values of these parts in your set? Do they match the values indicated in the schematic? Did you test them?

John Matheson
 
 
AUS  Articles: 4
Schem.:
Pict.: 0
30.Mar.21 06:58

Count of Thanks: 1
Reply  |  You aren't logged in. (Guest)   5

I pulled the filter components and they typically measured within 1% tolerance on both of my LCR meters. The worse item was C312 which measured 4% high. 

Looking again at the schematic I did miss the two sets of values notated on the circuit for C321 & C322 to set 50 or 75μS time constants depending on the market, so as you pointed out the cascaded filters R311-C311 and R313-C321 - R312-C312 and R314-C322 respectively do combine to provide approximately the required deemphasis. 

FM pre-emphasis is created by a single first order high pass filter response, and the correct deemphasis is achieved by a complementary single first order low pass filter. It is not correct to add the time constants of cascaded filters and assume that combined the two sections act as a single pole filter with an fc of the sum of the time constants.

The implementation of deemphasis in the 2216 does not conform to the single pole response, with fc's at ~6kHz and ~10kHz, instead of a single fc at ~3.2kHz.  The cascaded filter starts with a first order response followed about an octave higher by a second order response, so it is too shallow to start with and then too steep to finish off.

The output of the cascaded filters is above the correct deemphasis curve by ~3db in the 3-12kHz range, tapering off at either end. That is is why the audio from the tuner sounds ‘bright’. Ironically as built the tuner section meets its specification of +/- 2dB, the broad treble hump being accommodated within the specification tolerance.

To fix this, I removed C311 and C31, and shunted C321 and C322 with 4n7 to approximate the value 9nF and create a single first order deemphasis filter with TC= ~53μS. Result: tuner output measures flat with the same high frequency rolloff as my reference tuner, and no longer sounds bright.

Thanks again for your input. It encouraged me to get to the bottom of the issue and improve on the jerry-rigged solution of my first post.

This article was edited 30.Mar.21 09:45 by John Matheson .

  
rmXorg