GRAPHITE RESISTORS - Troels Gravesen

25, Aug. 2025

 

GRAPHITE RESISTORS - Troels Gravesen

Graphite Resistors
Copyright -16 © Troels Gravesen

If you are looking for more details, kindly visit Advanced Targets.

Make your own non-inductive graphite resistors

Wikipedia: The mineral graphite is one of the allotropes of carbon. It was named by Abraham Gottlob Werner in from the Greek ãñáöåéí (graphein): "to draw/write", for its use in pencils, where it is commonly called lead, as distinguished from the actual metallic element lead. Unlike diamond (another carbon allotrope), graphite is an electrical conductor, a semimetal, and can be used, for instance, in the electrodes of an arc lamp. Graphite holds the distinction of being the most stable form of carbon under standard conditions. Therefore, it is used in thermochemistry as the standard state for defining the heat of formation of carbon compounds. Graphite may be considered the highest grade of coal, just above anthracite and alternatively called meta-anthracite, although it is not normally used as fuel because it is hard to ignite.

Let me say already here: I do not find any sonic difference in performance from wire wound, graphite, MOX, Duelund and film resistors tested on my Jenzen D (diamond tweeter). The differences they may display in induction is infinitesimal and cannot count for any sonic difference. Their difference in resistance vs. temperature may vary and should be considered where high levels of current is passing. Buying seriously expensive resistors is a waste of money - to my experience. Clean woodoo. The problem here is that when we buy something expensive, it just got to be better.

The special feature of a graphite resistor, except for being non-inductive, is that it displays a negative temperature coefficient (a = -0.5*10^-3, (1/°C), thus when the voice coil heats up - and increase impedance - the graphite resistor will counteract this by lowering impedance. In practice this is most likely rubbish as the graphite resistor would have to act as an exact mirror of what is going on in the voice coil. This is a very unlikely scenario, but that's how the story goes and as always, no measurements have ever - to my knowledge - been published that supports the claim. Voice coils come in all forms with regard to former (alu, kapton, etc.), wire material, wire diameter and length, and to claim that any graphite resistor in front of this will "read" the voice coil and counteract fluctuations in impedance is obviously nonsense. I many cases the attenuation resistor(s) is/are not directly in front of the driver, rather in front of the crossover components, thus the crossover still sees a fluctuating impedance and we're nowhere. If we have a problem with over-heated voice coils, we should build bigger speakers.

Nevertheless, it is claimed that graphite resistors sound better compared to MOX or wire-wound, and if a graphite resistor will make you happy you should go for it, as it can be made for nothing from the graphite rod found in any standard pencil. We only need to know the grades of pencils to buy.
To find out about the resistance of pencil graphites, I bought a range of pencils: 2B, B, HB, H and 2H, where 2B is the softest of the grades. A google search comes up with this wider range:

Conversion of grades from UK to US looks like this: B = #1, HB = #2, F = #2½, H = #3 and 2H = #4 (Wikipedia again).

My pencils are 175 mm long and the first thing to do is to find out what's the resistance of the full-length graphite rod, thus the wood was removed at the upper end and resistance measured:
2B = 6 ohms, B = 7 ohms, HB = 19 ohms, H = 25 ohms and 2H = 20 ohms. Quite surprising actually. I thought we'd have a steadily increase in resistance vs. hardness.
So far so good. Apparently I only had to buy two grades to get rods that will cover the range of resistors needed for crossover work, that is 0.47 ohm to - let say 15 ohms. I rarely use resistors above 15 ohms for crossover work. The reason for choosing two grades is that the resistors otherwise will be long and not easy to physically implement on the crossover board. 8-12 cm length seems appropriate. Fortunately someone gave me a bag of Chinese made pencil rods (pic above to the right). Otherwise I suggest you simply burn a stack of pencils with your weed burner. Feel sure the graphite will survive. To cut off the wood is tedious and a lot of rods will break. I've tried this also!

Resistivity of pure graphite is * 10^-8 ohms/m/m^2, that is 0. ohm/meter. Now, 1 square meter of graphite is quite a lot, actually 1,000,000 mm^2, thus 1 mm^2 is 13.8 ohm/meter. The rod here being 2.2 mm diameter have a square area of 3.46 mm^2. So, one meter of pure graphite pencil should have a resistance of 3.99 ohms and a 17.5 cm pencil 0.7 ohms. Actually they're not, rather 6-25 ohms, thus some clay has been mixed in. How much we don't know and this is probably a secret for every pencil manufacturer and who cares? Fortunately an amount of clay has been added that provides us with something useful for crossover resistors with regard to physical length.

The grades of hardness are made by mixing the graphite with various amounts of clay powder. Next the rods are dipped in wax or oil to provide smoother writing. The latter we have to get rid of to make a proper termination and you may boil your graphite rods is caustic soda or simply burn the ends as seen below. I use our weed burner and it only takes a few seconds to heat the graphite until it glows. No wax or oil will be left after this!

Before you rush to your local pencil supplier, there's one important thing you need to make proper resistors: A low-ohm meter! These definitely do not come cheap, so in the end it may be cheaper to buy the commercially available graphite resistors. I paid ~400 USD for my meter. A standard meter may be used but make sure you have 1% resistors at hand as controls.

Thanks to Max, here's a link to a supplier (California) of pencil leads: http://www.jetpens.com/index.php/cPath/99

Another link here from Mark, http://www.suppliesnet.com/, (Louisiana, US) search for "leads" and you can find from 3B to 6H. The soft range will be interesting for low-ohm resistors as the leads I have will make quite short resistors for e.g. 0R47, 1R0, etc.

The pencil graphite rod is exactly 2.2 mm in diameter, and you need a 2.1 mm drill fixed in a vise as seen above. And I'm serious here. Not 2.2 or 2.0 mm but 2.1 mm. The reason for this is that you have to twist 0.8 mm copper wire 5 times around this 2.1 mm drill, making it a fraction smaller than the 2.2 mm graphite rod. This way you can screw the copper wire onto the graphite rod (counter-clock) and it stays tight. The copper wire has to be 0.8 mm, not 0.7 or 0.9 mm. I've tried a lot of wires - and 0.8 mm copper wire just works. You may use silver wire, but will have to find out the optimum diameter as silver is softer compared to copper and may work differently. And by the way: There will - usually - be loads of copper wire in your crossover components so to use silver here wire is only show-off.

To do some further experiments I needed more 0.8 mm copper wire and it turned out my local wire pusher was out on 0.8 mm wire. We searched countless shelves for some 0.8 mm wire, even the multi-strand installation cables. No 0.8 mm left could be found.
So, I bought some 0.8 mm silver wire (999) and this turned out to work absolutely perfect. Whether you need some high-silver solder is a good question. I tried both and it worked fine. 0.8 mm silver wire is not exactly cheap and each resistor will set you back 1-1.5 EUR depending on how long wires you need.

Next you need to remove the wax/oil from the graphite rods and as mentioned earlier, a weed burner works excellent. And DO NOT put your finger on a newly burned graphite rod. It's hot like hell and will burn you to your bones! Right photo: Copper wire screwed onto graphite rod.

Next you need to find out the length of the graphite rod to produce the desired resistance. Cut the rod at some 5 mm extra and repeat heat treatment. Right photo: Solder the first copper termination and heat it for a considerable length of time until the solder has fully penetrated all windings. This way a small amount of solder will go below the windings as the copper will expand during heating and once cooled - and shrunken, the termination will stay rock-solid and you shouldn't be able to pull it off without damaging the graphite. At least I can't. I have had the opportunity to check one commercial graphite resistor and can say that the termination shown here is just as good if not even better. Finally add the second termination and keep screwing until you reach target resistance and solder.

Left photo: Check resistance and cut off excessive graphite. Right photo: I also happen to have some pertinax tubes and they work excellent for housing the fragile graphite rods. Slice the tube at both ends and push the terminated graphite rod through and pull out the wires. Seal one end with Superfix or similar and fill the tube with fine dry sand for cooling. Seal with Superfix and you're done!
Alternatively you may use cardboard tubes or similar.

So, what's the watt rating of our homemade graphite resistors? Hard to tell, but at least it should do until the solder melts and I'll complete the study with some tests asap.

There's a good reason for the high price on commercially available graphite resistors. The development phase was great fun, but once you've made a few, it becomes boring and I would definitely charge you at least 20 USD for a resistor. And please do not ask, because I don't make these to order.

Appendix I

Burning pencils takes a little practice: The hard graphite, from HB to 2H is brittle and will fracture if subjected to uneven heat. Place the pencils on your garden grill with 1 cm apart. Heat gently until the wood catch fire and apply only as much heat as necessary to keep burning. Once the wood is gone, let it cool and clean the graphite rods with steel wool and heat the ends as shown above to prepare for termination.

Appendix II

Above the resistance of a HB rod vs. temperature. The graphite was cut at ~15 ohms and terminated with teflon wires to allow the resistor to be placed in our kitchen owen. Temperature was checked with a thermometer and we're doing +/- 3 ºC of target. Not bad for a kitchen owen! What we see is a nice linear decline in resistance vs. temperature. An increase in temperature of 230 ºC will lower the resistance ~18%. The resistance was not corrected for the increase in resistance of the teflon wires used for measurements as these are close to zero, thus will only have a minor impact on results.
For comparison a softer B rod was tested and had a decline of 13% from 20-200 ºC; quite similar to the HB rod.

Target Variables - Graphite Note

Target variables are a significantly important concept in predictive modeling and machine learning. Target variables have a direct effect on the accuracy and effectiveness of machine learning models.

Defining target variables

A target variable is also known as a dependent variable. A target variable is the outcome you aim to predict or explain using your machine learning model. A target variable is the variable that you want to estimate or classify based on the available data.

If you are looking for more details, kindly visit graphite target.

Target variables in machine learning

Target variables guide the machine learning process. Target variables provide a benchmark for your machine learning model’s performance. You can assess the accuracy and effectiveness of your model by comparing the predicted values to the actual values of the target variable. Target variables serve as the basis for model training. By exposing the model to a large dataset with known target values, the model can learn patterns and relationships.

This enables your machine learning model to make accurate predictions or classifications when faced with unseen data. Choosing your target variable is key. Choosing your target variable determines the type of problem you are trying to solve. It also determines the appropriate algorithms and techniques to use. Different types of target variables require different approaches and considerations.

Different Types of Target Variables

Target variables can take various forms depending on the nature of the problem. These are the different types of target variables:

Categorical variables: Categorical variables represent distinct classes or categories. Categorical variables are used in classification problems. These include predicting whether an is spam or not. Categorical variables can have two or more classes. Your machine learning model must assign the correct class to each instance based on the available features.

Numerical variables: Numerical variables take on continuous values and are usually used in regression problems. These include stock price predictions, or similar. Your machine learning model must estimate the numerical value of the target variable based on the input features and the patterns observed in the input data and training data.

Ordinal variables: Ordinal variables have a specific order or rank. Ordinal variables are used to solve problems where responses are rated on a scale. Your machine learning model must predict the ordinal value or rank of the target variable based on the available features.

Understanding the nature of your target variable influences how you select your machine learning model. Your target variable affects your machine learning model’s accuracy.

Target variables in predictive modeling

In predictive modeling, target variables are important for enhancing a machine learning model’s accuracy. Target variables are the benchmark for evaluating the performance of your machine learning model. You compare the predicted values to the actual values of the target variable. This assessment tells you about your machine learning model’s accuracy. This assessment also helps you identify areas for improvement, enabling higher levels of accuracy and precision. Target variables also give you insight into the underlying patterns and relationships in the data.

Target variables and accuracy

A predictive model can be used to predict customer churn in a subscription-based service. The target variable would be whether a customer churns or not. By analyzing the relationship between the target variable and customer attributes, we can identify the key factors that contribute to customer churn. This understanding can then be used to implement targeted retention strategies and reduce customer attrition

The effect of incorrect target variables

Picking the wrong target variable messes up your predictions. If it’s not what you’re trying to predict, the model won’t work.  If you predict income (a number) as low/medium/high, the model misses details and can’t predict as well. Choosing a clear target variable is key to getting good results from your machine learning model.

How to choose the right target variable

Choosing the right target variable requires careful consideration of several factors. Here are some key points to keep in mind:

  • Relevance: Your target variable should be directly related to the problem you are trying to solve. Your target variables should reflect the information or outcome you want to predict or classify.
  • Availability: Ensure that you have a sufficient amount of data available that have known target values. A target value is a key metric for you. Without a substantial dataset, your model may struggle to learn meaningful patterns.
  • Measurability: The target variable should be measurable. Your target variable should be something that can be classified objectively.
  • Balanced Distribution: When you have a categorical target variable, aim for a balanced distribution among the classes. This ensures that your machine learning model is not biased towards a particular outcome

Common mistakes when choosing target variables

Choosing the wrong target variable affects your machine learning model’s accuracy and effectiveness. Avoid these common mistakes when choosing your target variables:

  • Choosing an irrelevant target variable that does not provide meaningful insights or predictions.
  • Mistaking a derived variable as the target. This leads to circular reasoning and flawed results.
  • Ignoring the relationship between the target variable and the input variables. You will miss out on valuable information

Pre-processing target variables

You need to pre-process your target variables before inputting them into your machine learning models. You should handle missing target variables for your predictve model with:

  • Imputation: Imputation estimates missing values based on the available data. Common imputation methods include mean imputation and multiple imputation.
  • Exclusion: If you have significant amounts of missing data, you may need to exclude those instances from the analysis. Be careful here, to ensure that your data analysis remains representative and unbiased.
  • Normalizing: Normalization ensures that the target variable lies within a specific range. This makes it easier for your machine learning model to learn and make accurate predictions.
  • Scaling: Scaling adjusts the variance of your target variable. Scalin is useful when dealing with machine learning models that are sensitive to variable scales.

Assessing the performance of target variables in predictive modeling

Assessing the effectiveness of your target variable depends on the type of problem you are trying to solve:

  • Classification problems: Accuracy, precision, recall, and F1-score are used metrics to measure the performance of target variables in classification problems.
  • Regression Problems: Mean absolute error (MAE), mean squared error (MSE), and R-squared are used to measure the effectiveness of target variables in regression problems.

Improve target variable performance

If you need to improve your target variable’s performance, you can use:

  • Feature engineering: Create new derived features from existing data. That may enable you to discover additional predictive power in the target variable.
  • Data augmentation: Increase the size and variety of the dataset to expose your machine learning model  to a wider range of patterns.
  • Model selection and optimization: Experiment with different machine models and fine-tune your hyperparameters

So, there you have it! A comprehensive understanding of target variables in machine learning. Remember, the choice and proper preprocessing of a target variable are critical for accurate predictions and effective model performance.

Graphite Note is here to help you harness the power of machine learning without the need for complex coding. Whether you’re part of a growth-focused team, an agency without a data science department, or a data analyst looking to delve into AI, our platform is designed to transform your data into precise predictions and actionable strategies with just a few clicks. Experience the synergy of our no-code predictive analytics tools and see how easy it is to predict business outcomes and make data-driven decisions.

Request a Demo today and take the first step towards unlocking unparalleled insights and efficiency for your business.

Transform your business with Graphite Note’s prebuilt machine learning models that work with any dataset, regardless of industry. Our data-agnostic platform delivers insights and predictions in minutes, giving you a competitive edge. Explore high-impact use cases that drive real business outcomes:

Key Use Cases:

1. Lead Scoring: Prioritize high-quality leads and increase conversion rates. Customer Churn Prevention: Retain more customers by identifying those at risk of churning.

2. Cross-Sell Optimization: Boost sales by targeting customers likely to purchase additional products.

3. Time Series Forecasting: Optimize inventory and plan for sales peaks and low periods.

4. Customer & Product Segmentation: Segment customers for personalized strategies and sales growth.

5. Customer Lifetime Value: Predict customer value over time to maximize revenue.

6. Predictive Maintenance: Reduce downtime, cut costs, and extend machinery life. Start leveraging Graphite Note today for smarter decisions and faster results!

For more titanium sputtering targetsinformation, please contact us. We will provide professional answers.