Output Volume Differences between CPU Revisions

Discuss technical or other issues relating to programming the Nintendo Entertainment System, Famicom, or compatible systems. See the NESdev wiki for more information.

Moderator: Moderators

Post Reply
Great Hierophant
Posts: 780
Joined: Tue Nov 23, 2004 9:35 pm

Output Volume Differences between CPU Revisions

Post by Great Hierophant »

I have been interested in testing the differences in loudness between the various revisions of the NES CPU. The conventional wisdom is that earlier Famicoms' internal sound is louder than the later Famicoms or the AV Famicom. I decided to compare the output of my Famicom and AV Famicom. However, the default audio output of the Famicom is through the RF modulator while the AV Famicom outputs an untouched audio signal.

The CPU in my Famicom is a Revision E and the board is a common HVC-CPU-07, the CPU in my AV Famicom is a Revision H (laser etched) and the board is an HVCN-CPU-02. In order to control for all variables, I decided to tap the audio from the Famicom Expansion Port. The port is very convenient because Pin 1 is Ground and Pin 2 is Sound Out. The port only carries internal audio, you won't hear cartridge expansion audio from it. As far as I can tell, the audio mixing and amplification circuitry is identical on all Famicoms and AV Famicoms at the point they are passed to the Famicom Expansion Port, but I could be proven wrong. I know at least there are two 100Ohm, a 20K, a 12K and a 1uF 50V cap common to all Nintendo consoles and they all pass their audio through an inverter.

I recorded the opening tune from Metroid running on my Famicom and AV Famicom. I made sure to record at the same session and at the same volume level (50%) from my PC's line input. Here are my recordings, preserved in the best quality :

http://www.mediafire.com/file/gm71kdx93 ... etfam.flac

http://www.mediafire.com/file/d2cocqxqm ... famav.flac

It turned out that there was no real appreciable difference in the volume whether I recorded from the Famicom Expansion Port or the Multi-Out of the AV Famicom. At the volume level I recorded at, there was an audible noise floor from the Famicom that was not present on the AV Famicom. The Famicom is noticeably louder, perhaps by 4.5db. This means that expansion audio in cartridges will sound quieter on an older Famicom compared to an AV Famicom.

However, I also made a recording of Metroid's opening on my front loader NES, which has a Revision G CPU, through its RCA Audio Output connector. The volume was lower than either Famicom, which suggests that I should really try to record a later Famicom with a Revision G CPU (The VCI marked Famicoms with the RF board soldered to the main PCB).

http://www.mediafire.com/file/da9i8or9k ... etnes.flac

I don't know why the NES is quieter than the AV Famicom, but I note there seems to be a lot more going on with the NES audio's output than the AV Famicom's. I suspect that if I tapped the audio at the NES Expansion Port, I would have louder audio, but I don't have an ENIO board and I don't have another spare cable to sacrifice by soldering it to the pins.
User avatar
rainwarrior
Posts: 8732
Joined: Sun Jan 22, 2012 12:03 pm
Location: Canada
Contact:

Re: Output Volume Differences between CPU Revisions

Post by rainwarrior »

Great Hierophant wrote:As far as I can tell, the audio mixing and amplification circuitry is identical on all Famicoms and AV Famicoms at the point they are passed to the Famicom Expansion Port, but I could be proven wrong. I know at least there are two 100Ohm, a 20K, a 12K and a 1uF 50V cap common to all Nintendo consoles and they all pass their audio through an inverter.
I wouldn't expect the mixing to be "identical". Resistors, for instance, are only manufactured to within a specified tolerance of the marked value (e.g. 10%); unless they're using expensive high precision components, there's a lot of range here.

I don't truthfully know how much variance there will be between CPU revisions, but the variance from mixing components is already pretty high, I think, so I don't think you're really going to get a good bead on the CPU itself this way. (Maybe if you could measure directly from the CPU chip, but you'd probably want to remove them from the board and use one consistent power supply too, if you really wanted to know that?)

Whether or not the data has much to do with the isolated CPU effect, though, it'd be good to get at least some relative level measurements for the 2 APU channels surveyed across a wide assortment of NES machines, just to get a better understanding of what the average balance is overall.
lidnariq
Posts: 11432
Joined: Sun Apr 13, 2008 11:12 am

Re: Output Volume Differences between CPU Revisions

Post by lidnariq »

It is conceivable that at least one of the CPU revisions also included a silicon process revision, perhaps going from 6µ to 4µ or changing the oxide thickness. I don't know if our current photos (letterless and G revisions) have their scale documented.

Another possibility is that analog behavior is extremely specific to the specific process being used (there are many anecdotes of mixed-signal ICs being discontinued specifically because of the old fab being shut down), and so if the same masks were used at different locations that might make the difference? (i.e. maybe it's not "E" vs "H" but country A vs country B)
Great Hierophant
Posts: 780
Joined: Tue Nov 23, 2004 9:35 pm

Re: Output Volume Differences between CPU Revisions

Post by Great Hierophant »

rainwarrior wrote:
Great Hierophant wrote:As far as I can tell, the audio mixing and amplification circuitry is identical on all Famicoms and AV Famicoms at the point they are passed to the Famicom Expansion Port, but I could be proven wrong. I know at least there are two 100Ohm, a 20K, a 12K and a 1uF 50V cap common to all Nintendo consoles and they all pass their audio through an inverter.
I wouldn't expect the mixing to be "identical". Resistors, for instance, are only manufactured to within a specified tolerance of the marked value (e.g. 10%); unless they're using expensive high precision components, there's a lot of range here.

I don't truthfully know how much variance there will be between CPU revisions, but the variance from mixing components is already pretty high, I think, so I don't think you're really going to get a good bead on the CPU itself this way. (Maybe if you could measure directly from the CPU chip, but you'd probably want to remove them from the board and use one consistent power supply too, if you really wanted to know that?)

Whether or not the data has much to do with the isolated CPU effect, though, it'd be good to get at least some relative level measurements for the 2 APU channels surveyed across a wide assortment of NES machines, just to get a better understanding of what the average balance is overall.
Nintendo used resistors with a gold band throughout their Famicoms, indicating 5% tolerance. Not exactly high end for consumer electronics, but still standard enough that they need not hold their heads in shame.

I have found that the "group of four" APU pin mixing resistors use the same nominal resistances throughout the Famicoms and the Front Loader NES. The Top Loader NES uses 200 and 160 Ohm values in place of 20K and 12K Ohms. The triangle, noise and DPCM may be a little louder on these systems.
User avatar
rainwarrior
Posts: 8732
Joined: Sun Jan 22, 2012 12:03 pm
Location: Canada
Contact:

Re: Output Volume Differences between CPU Revisions

Post by rainwarrior »

Another thing to look out for is heat. I've definitely seen some variance in balance/volume between just starting up and after running for a while, especially w.r.t. expansion audio.
User avatar
nothingtosay
Posts: 39
Joined: Mon May 19, 2014 11:46 pm

Re: Output Volume Differences between CPU Revisions

Post by nothingtosay »

I have an HVC-CPU-07 Famicom and an HVC-CPU-GPM-02 Famicom and I made test recordings recently of Metroid to check out the differences I'd heard of. Here are comparison files I already had uploaded to share with someone else. http://www.mediafire.com/file/3sg2fgl78 ... arison.zip

The one marked OG Famicom is the CPU-07 and FF (for Famicom Family, the logo this model has on the front to the left of the FAMILY COMPUTER label) is the GPM-02. The main comparison is of recordings of the full Metroid title screen music (what a coincidence), but there's also a short clip of "Brinstar" to demonstrate briefly the difference with some music that doesn't use the FDS expansion audio. This was recorded through a 3-foot RG6 coaxial cable with an RCA adapter plugged into the back of the Famicoms going into this RF demodulator with HDMI conversion and output, run into an HDMI splitter with the audio captured bit-perfectly over S/PDIF into my M-Audio 2496 sound card. I used the same coaxial cable, power supply, and RAM adapter with FDSStick for both Famicoms with the RF demodulator set to the same volume level. I simply recorded the first one, then moved all the stuff over to the second one and recorded it. I also boosted the volume equally for both after recording for ease of use since it was pretty quiet. It was over a month ago I made these recordings, but I don't think I had either system running for long before recording so I expect heat conditions to be about the same.

(Unfortunately, the CPU-07 Famicom has since blown a capacitor or something so I'm not sure I can make a re-recording where I can be certain temperatures are equal. I use an original power supply with a step-down transformer and it worked for many hours of use without issue, but last time I tried to use it there was a popping sound and white smoke came out of the vent at the back, which was very disconcerting and I don't understand what caused it. :cry: I opened it up and looked at the board but didn't see anything obviously wrong, but I'm not trained or knowledgeable in these things and I haven't tried to use it again for fear of making it worse.)

My result, as you can see if you look at the waveforms and listen to them, is that the output volume of the 2A03 audio is essentially the same between these examples of the two models. However, the GPM-02's sound is a bit less trebly and you can hear this pretty well in the short comparison clip of just the 2A03 square wave audio. I'd read that 2A03 output volume is lower on the Famicom Family models, but that appears not to be the case, aside from losing that bit of treble. Instead, the expansion audio on the GPM-02 is significantly louder.
Great Hierophant
Posts: 780
Joined: Tue Nov 23, 2004 9:35 pm

Re: Output Volume Differences between CPU Revisions

Post by Great Hierophant »

nothingtosay wrote:I have an HVC-CPU-07 Famicom and an HVC-CPU-GPM-02 Famicom and I made test recordings recently of Metroid to check out the differences I'd heard of. Here are comparison files I already had uploaded to share with someone else. http://www.mediafire.com/file/3sg2fgl78 ... arison.zip

The one marked OG Famicom is the CPU-07 and FF (for Famicom Family, the logo this model has on the front to the left of the FAMILY COMPUTER label) is the GPM-02. The main comparison is of recordings of the full Metroid title screen music (what a coincidence), but there's also a short clip of "Brinstar" to demonstrate briefly the difference with some music that doesn't use the FDS expansion audio. This was recorded through a 3-foot RG6 coaxial cable with an RCA adapter plugged into the back of the Famicoms going into this RF demodulator with HDMI conversion and output, run into an HDMI splitter with the audio captured bit-perfectly over S/PDIF into my M-Audio 2496 sound card. I used the same coaxial cable, power supply, and RAM adapter with FDSStick for both Famicoms with the RF demodulator set to the same volume level. I simply recorded the first one, then moved all the stuff over to the second one and recorded it. I also boosted the volume equally for both after recording for ease of use since it was pretty quiet. It was over a month ago I made these recordings, but I don't think I had either system running for long before recording so I expect heat conditions to be about the same.

(Unfortunately, the CPU-07 Famicom has since blown a capacitor or something so I'm not sure I can make a re-recording where I can be certain temperatures are equal. I use an original power supply with a step-down transformer and it worked for many hours of use without issue, but last time I tried to use it there was a popping sound and white smoke came out of the vent at the back, which was very disconcerting and I don't understand what caused it. :cry: I opened it up and looked at the board but didn't see anything obviously wrong, but I'm not trained or knowledgeable in these things and I haven't tried to use it again for fear of making it worse.)

My result, as you can see if you look at the waveforms and listen to them, is that the output volume of the 2A03 audio is essentially the same between these examples of the two models. However, the GPM-02's sound is a bit less trebly and you can hear this pretty well in the short comparison clip of just the 2A03 square wave audio. I'd read that 2A03 output volume is lower on the Famicom Family models, but that appears not to be the case, aside from losing that bit of treble. Instead, the expansion audio on the GPM-02 is significantly louder.
If you got white smoke, it was probably one of the caps or the 7805 voltage regulator. Since the HVC Famicoms only have four caps, you should probably replace them and the voltage regulator. Find a replacement cap kit here : https://console5.com/store/famicom-cap- ... c-001.html

As far as the internal audio goes, how can one console make expansion audio sound quieter than the other console when the volume of the internal audio is the same? I think I need to get my own GPM Famicom.
Post Reply