Dimehead NAM Player

[MirrorProfiles]

1000 hz sine wav, blue is pedal, yellow is loopback, -9.71 dbFS (according to rew which is generating the signal). It's measuring exactly 1 volt at the tip into pedal. Putting it up a hair a above this will start flattening the peaks (which is exactly where it hits the top of the meter on the pedal). Internal gain is at 5 but there's no nam file loaded

View attachment 28956

a 1V sine wave into a 12dBu input would measure about -9.71dBFS, so presuming that’s where the other DI is going as the numbers match up exactly?

And if the the Dimehead is clipping at 1V, that actually makes it seem like the headroom is slightly lower than 4dBu?

You can think of it like any sample recorded higher than -9.71dBFS in a 12dBu input would get clipped on the Dime (or there is 9.71dB less headroom than a 12dBu input).

 

[MirrorProfiles]

I guess the other thing to check would be whether lowering the gain knob in the pedal lowers the clipping or just the output. Presumably the signal will stay clipped despite the level changing.

You could probably work out the scaling of the gain knob though, by noting the difference in output when the gain is at different values.

 

[Dom Mcsweeney]

The loudest peaks of this DI are all at the exact same level, so definitely clipping the A/D.

View attachment 28950

example above kind of shows - the peaks are getting taken off the DI before the processing has a chance to do anything. Ideally the signal that is processed is as close to possible as what left the guitar, rather than one that has had the transients cut off from the converters being overloaded.

Hard to verify whether that DI has 4dBu of headroom or more by looking at a waveform though, all that it is certain is the headroom of the inputs is lower than the output of the pickups.

that's what I said... it's me playing guitar as loud as hell, at full gain
only if i do that do I get any sort clipping

i.e. I have to be playing stupidly heavy chugs to force it to happen, and so my 'chugs' are basically my own 'full gain' and the unit is at full gain


so just like with an audio interface, you probably wouldn't have it at full gain..... unless you want that to happen

 

[northern_fox]

That's almost like defying physics.
Signals nulling perfectly are bit for bit identical. Hard to imagine when one was running through an ADDA chain and at least some kind of op-amp.

Yes, it was low enough that I had to crank the levels to hear it, but not absolutely zero

 

[northern_fox]

I guess the other thing to check would be whether lowering the gain knob in the pedal lowers the clipping or just the output. Presumably the signal will stay clipped despite the level changing.

You could probably work out the scaling of the gain knob though, by noting the difference in output when the gain is at different values.

Gain knob only affects the output. From dirk:

Gain Poti 0 - 10 corresponds to amplification of 0-x5 with a log poti curve. Position 2.4 corresponds to a gain of x0.4 = -8dB

 

[MirrorProfiles]

that's what I said... it's me playing guitar as loud as hell, at full gain
only if i do that do I get any sort clipping

i.e. I have to be playing stupidly heavy chugs to force it to happen, and so my 'chugs' are basically my own 'full gain' and the unit is at full gain


so just like with an audio interface, you probably wouldn't have it at full gain..... unless you want that to happen

If 1V is clipping, then the maximum input level the pedal can accept before clipping is 2.2dBu (aka 1V RMS). A guitar can output over 3V RMS, so 1V of headroom is not ideal. The same is true if you run a pedal in front of the Dime - your signal will very quickly exceed 1V. Something like a Dirty Tree can add a LOT of level far beyond 3V.

It will clip with loud pickups + playing, yes. I can JUST about clip 12 or 13dBu inputs with loud playing but it’ll only be knocking off a bit of the loudest peaks. Those same signals through a Dime would be getting 10dB clipped off which is not really adequate headroom. It would be better to pad the input down 10dB or so and then gain up in the DSP.

 

[northern_fox]

1 volt for a 1 khz sine wave. I believe it'll be different depending on that frequency

For fun, here's a test with varying frequencies and input levels, and you can see where clip starts. (turn your volume down first)

 

[northern_fox]

Interesting - here's the same test for tonex. The input trim doesn't have any effect on the preamp either (I checked to make sure it was the input clipping and not the output)! It's a post preamp trim. The headroom here is significantly higher. The clip point for the tonex was ~ 2V vs 1V on the dimehead.



Here's a side by side of the two with the same signal: The vertical divisions are not equal in this video, apologies for the quality.

 

[MirrorProfiles]

Interesting - here's the same test for tonex. The input trim doesn't have any effect on the preamp either (I checked to make sure it was the input clipping and not the output)! It's a post preamp trim. The headroom here is significantly higher. The clip point for the tonex was ~ 2V vs 1V on the dimehead.



Here's a side by side of the two with the same signal: The vertical divisions are not equal in this video, apologies for the quality.

(If you haven’t already) can you check using the ToneX pedal as an interface to record DI’s and to see the headroom that way?

Thanks so much for doing these, very helpful and interesting to see.

2V RMS=8.2dBu of headroom.

 

[northern_fox]

(If you haven’t already) can you check using the ToneX pedal as an interface to record DI’s and to see the headroom that way?

Thanks so much for doing these, very helpful and interesting to see.

2V RMS=8.2dBu of headroom.

I don't have a way to clip it with pickups, so I'm not sure how I could establish the headroom that way (besides reamping a DI into it and boosting the level, which I think is effectively the same test that I've just done with the sine wav?)

 

[MirrorProfiles]

I don't have a way to clip it with pickups, so I'm not sure how I could establish the headroom that way (besides reamping a DI into it and boosting the level, which I think is effectively the same test that I've just done with the sine wav?)

Oh I wasn’t sure if you were using ToneX with no model loaded, or as an interface. You could do the same test with the sine wave (assuming it’s possible to run ToneX without a model loaded, I’ve never used the HW pedal).

 

[northern_fox]

Oh I wasn’t sure if you were using ToneX with no model loaded, or as an interface. You could do the same test with the sine wave (assuming it’s possible to run ToneX without a model loaded, I’ve never used the HW pedal).

Ah sorry - no model loaded, no cab, all fx disabled, all controls flat - straight through. Not in interface mode, just in->out the same as dimehead

 

[[Nathan]]

My enjoyment of plugins immediately changed when you posted that Google spreadsheet a year or two ago.

The only reason I ever paid attention to this topic is because I try out almost all amp sims and so often they were hit or miss yet the product demos always sounded amazing. I could tell it wasn’t just mixing tricks it was like something was fundamentally different. Low and behold many of these ones that sounded bad just needed a savage boost…. And it’s not a guess it’s literally the dev saying use this number to experience the software as we do.

I don’t have the time or mental energy to flick through nam profiles that don’t have an input spec with it, heck even my own old profiles I don’t bother loading up because I probably used different reamp values. I have a jhs bonsai boost pack at 13dbu and I use that all the time with different amp sims and it always works like a champ (because the results are repeatable and consistent). I also grabbed 2dors recent King Kong pack and it sounded perfect in 2 seconds because of this info.

With a proper software guided capture and calibration process you could bury this input metadata into nam models and then software/hardware would just auto adjust things so we never have to think about this stuff again. But downplaying its existence or critical role is only going to hold things back. I’m sure tons of people have tried nam with mixed results and have just gone, I’ll stick to neuraldsp cause it sounds good with no effort. I’ve always said the nam tech is there it just needs to be harnessed and streamlined for ease of use for a wider audience.

 

[[Nathan]]

I grabbed some info from the spreadsheet and added some other ones. +4 seems oddly low stacked up to everything.
11.5 - Line 6 Helix
16 - Fractal FM-3
15 - Neural DSP Quad Cortex
10 - Neural DSP Nano Cortex
10.5 - IK Axe I/O One
11.2 - ToneMaster Pro
4 - Dimehead NAM Player

My train of thought might be wrong here but I grabbed a couple guitars and played a riff at +4dbu and +10dbu, then I lowered the +4dbu down -6db so they would "match" and we can compare the waveforms (not scientific but ballpark).

Balaguer stock pickups

Fishman Fluence Modern (these seem to internally clip around +9dbu)

Bareknuckle Aftermaths

 

[MirrorProfiles]

I grabbed some info from the spreadsheet and added some other ones. +4 seems oddly low stacked up to everything.
11.5 - Line 6 Helix
16 - Fractal FM-3
15 - Neural DSP Quad Cortex
10 - Neural DSP Nano Cortex
10.5 - IK Axe I/O One
11.2 - ToneMaster Pro
4 - Dimehead NAM Player

My train of thought might be wrong here but I grabbed a couple guitars and played a riff at +4dbu and +10dbu, then I lowered the +4dbu down -6db so they would "match" and we can compare the waveforms (not scientific but ballpark).

Balaguer stock pickups
View attachment 29053

Fishman Fluence Modern (these seem to internally clip around +9dbu)
View attachment 29054

Bareknuckle Aftermaths
View attachment 29055

Nice work! and thanks for those values, will add to the spreadsheet.

DimeHead Player isn’t 4dBu though, it’s 2.2dBu (1V RMS=0dBFS). Even at 4dBu of headroom though it’s easy to see it’s not enough for humbuckers.

 

[northern_fox]

Nice work! and thanks for those values, will add to the spreadsheet.

DimeHead Player isn’t 4dBu though, it’s 2.2dBu (1V RMS=0dBFS). Even at 4dBu of headroom though it’s easy to see it’s not enough for humbuckers.

According to Dimehead / the spec it's 4 - just keep in mind this 1V is what I happened to measure here with a voltmeter and a sine wav. I'd probably go with what the official spec is if you're making this sheet (as all the others are)

 

[Dirk Dimehead]

Hey everyone, it's Dirk from DImehead chiming in

In the first 25 units, we had a max of 2.2dBu, but starting from the second batch with the introduction of the FX loop, we changed it to 4dBu.

For future units, we will increase it to 8dBu. This is a value that is not only used by ToneX but also by many high-end products like Strymon.

Excessively high headroom isn’t always an advantage. It can result in under-utilizing the AD converter range and, especially with high-gain amps, lead to unnecessarily high noise levels.

So far, all users have been satisfied with this setup. Single coils typically put out a max of 1Vpp, and medium humbuckers around 2Vpp. For active super high-output pickups, the max level could indeed be too low in some cases.

The left bar graph displays the AD converter levels up to 0dBFS.

Additionally, the presets behave the same in terms of gain across all revisions. The headroom is internally compensated.

As a reference, Kemper’s 5dBu has been completely fine for its thousands of users over the past 10 years as well.

 

[Dom Mcsweeney]

@Dirk Dimehead
Thanks for responding Dirk, good to see your participation here!

As I said previously, with my unit (first batch).
I don't get any clipping from my humbuckers unless I'm forcing it by setting the input gain to max, or close to it.
Eg. When gain is set to 8.0 there's no clipping, but set it to 9 or higher and I get a tiny amount of clipping when I'm playing purposefully louder than I would naturally, just for the sake of 'getting clipping'.
(That's on a blank preset)

In practise, I tend to have the input gain set to about 2.5 mostly, and that allows me to use that gain knob creatively for more drive if I want.

Whilst I'm not finding any issues personally with 2.2dbu, the later batches getting more headroom is definitely a good thing for anyone who requires that extra room.

Glad that's cleared up a little =)

 

[pipelineaudio]

To be fair, the line 6 g30 and g50 distort under moderately high output hum buckers as well and. Ivory seems to complain about that.

 

[MirrorProfiles]

Thanks for post @Dirk Dimehead , cool to have you posting here!

For future units, we will increase it to 8dBu. This is a value that is not only used by ToneX but also by many high-end products like Strymon.

IMO, this is still too low. Fractal use at least 16dBu, NDSP QC is 15dBu, Helix is 11.5dBu, TMP 11.2dBu. While Kemper ships with 5dBu, it’s adjustable - so if 5dBu is a problem, like it was for me, it’s at least possible to increase. A typical 9V pedal clips internally at around 12dBu, and it’s a common value used as a maximum on interfaces (from Focusrite/UA/Audient and others) and IMO is a more sensible value, as it’ll ensure more consistent results when units are using money on a computer rig (where they’ll be auditioning sounds), and when using the pedal. When auditioning sounds on their computer and interface, they’ll presumably be setting things up to avoid clipping. So I think some kind of estimated parity with a typical users setup is the logical way to go.

For active super high-output pickups, the max level could indeed be too low in some cases.

My Les Paul from 1990 has a stock pickup and clips 12dBu inputs. Similarly, a lot of users will be running pedals into their Dime Head as it’s designed to go on a pedalboard. Some clipping might be unavoidable but I think avoiding it is best. 2V peak to peak is 0.707V RMS, this sounds far too low for humbuckers. 3V RMS is a more common value to use for this.

Additionally, the presets behave the same in terms of gain across all revisions. The headroom is internally compensated.

So if the maximum input level (of some models) is 1V (2.2dBu), is it getting adjusted internally to 4dBu? Or are the 4 and 8dBu input versions being matched to 2.2dBu (when gain is at 5)? Either way, IMO, there probably needs to be a simple input adjustment stage where users can account for the input level in 0.1dB steps, rather than using the gain pot. That would make it easier to achieve unity when a user is using NAM models on a computer and when they want to achieve the same sound on the pedal.

If there is a seperate input calibration level, then the user can set the gain control on the NAM block to align with how the NAM plugin operates on their computer rig.

As a reference, Kemper’s 5dBu has been completely fine for its thousands of users over the past 10 years as well.

Most users had no idea their DI’s were getting obliterated, but luckily, they have a parameter to adjust this (which many users did). In fact, Kemper actually temporarily adjusted this for more headroom on an update.

I don't get any clipping from my humbuckers unless I'm forcing it by setting the input gain to max, or close to it.
Eg. When gain is set to 8.0 there's no clipping, but set it to 9 or higher and I get a tiny amount of clipping when I'm playing purposefully louder than I would naturally, just for the sake of 'getting clipping'.
(That's on a blank preset)

@northern_fox ’s examples demonstrate that the clipping occurs before the processing. So once your signal hits the A/D, it doesn’t matter where the gain control is set - the clipping has already occurred. Setting the gain control lower won’t increase headroom, this simply isn’t how it’s intended to work.

Also check Nathan’s examples above. Even at 4dBu several of his guitars are clipped on almost every note. It’s the same for me - I can occasionally clip 12dBu without “trying” to. 10dBu would be clipping quite often, 8dBu, almost all the time. There is a reason why most interface and HW manufacturers land on 3-4V as a sweet spot. It’s the sweet spot between noisefloor and headroom. The noisefloor argument is valid but only if it is a bigger concern than clipping. Ideally both aspects are well engineered so neither present an issue.